def elapsed(self, last_step):
"""
Return the full and incremental elasped time.
"""
total = (time.time() - self.start_time)
last = (time.time() - last_step)
return 'Total time: %s | Last step: %s' % (self.get_time(round(total,4)), self.get_time(round(last,8)))
def wrappedFunc(*args, **kwargs):
start_time = time.time()
val = func(*args, **kwargs)
end_time = time.time()
print '\n-------------'
print 'Function-name:', func.func_name
print 'Time: %fs' % (end_time - start_time)
print '-------------'
return val
def _wrapper(x, y, color, *args, **kwargs):
time_before = time.time()
func_call_result = func(x, y, color, *args, **kwargs)
time_after = time.time()
total_time_spend = time_after - time_before
print "Total time spent: {0}".format(str(total_time_spend))
print "Total array size: {0}\n".format(numpy.size(func_call_result.pixels))
timing_results.update(func_call_result(numpy.size(func_call_result.pixels), total_time_spend))
return func_call_result
def run(func, im, format, t):
global image
image = im
start = time.time()
set = t.repeat(iterations,1)
elapsed = (time.time() - start)
min_ = min(set)*1000
avg = (sum(set)/len(set))*1000
name = func.__name__ + ' ' + format
results[name] = [min_,avg,elapsed*1000,name,len(func())]
sortable[name] = [min_]
def show(self,**kwargs):
timeit = kwargs.get('timeit')
if timeit:
start = time.time()
# todo - we should check for postgis support, and otherwise
# use the MemoryDatasource
lyr_adapter = PostgisLayer(self)
lyr_adapter.show(**kwargs)
if timeit:
elapsed = (time.time() - start)/60
print 'render time: %s seconds' % elapsed
def output_time(self, print_time):
"""
Timing output wrapper to control the start point and verbosity of timing output.
"""
if self.timing_started and print_time:
val = color_text(4,self.elapsed(time.time()),self.no_color)
sys.stderr.write('%s\n' % val)
def total_time(self, last_step=None):
if self.verbose:
total = time.time() - self.start_time
out = "Total Nik2img run time: %s" % (self.get_time(round(total, 4)))
if last_step:
out += "| Last step: %s" % self.get_time(round(last_step, 8))
val = color_text(4, out, self.no_color)
sys.stderr.write("%s\n" % val)
def main():
frames = video_by_frame("./data/video.mp4")
tracking_frames = track_video(frames)
start_time = time.time()
for im in tracking_frames:
stop_time = time.time()
time_elapsed = 1 / (stop_time - start_time)
draw_str(im, (20, 20), 'FPS: %.2f' % time_elapsed)
start_time = stop_time
cv2.imshow('LKtrack', im)
key = cv2.waitKey(40)
if key == 27 or key == ord('q') or key == ord('Q'):
break
cv2.destroyAllWindows()
def generate_cache(path, files):
cache_dict = {}
time_zero = time.time()
for file in files:
start_time = time.time()
cprint('\n======= processing file {}...'.format(file), color='blue')
f = open(os.path.join(path, file), 'rU', encoding='utf-8')
raw = f.read()
f.close()
tokens = nltk.word_tokenize(raw)
book = nltk.Text(tokens)
hapaxes = len(nltk.probability.FreqDist(
book).hapaxes()) / len(set(book))
vocab_count = len(set(book))
print(
'=== {:.2f}s -- Text class initialized'.format(time.time() - start_time))
bigrams = nltk.bigrams(book)
bi_cfd = nltk.ConditionalFreqDist(bigrams)
print('=== {:.2f}s -- bigrams done'.format(time.time() - start_time))
trigrams = nltk.trigrams(book)
tri_cfd = nltk.ConditionalFreqDist(((x, y), z) for x, y, z in trigrams)
print('=== {:.2f}s -- trigrams done'.format(time.time() - start_time))
file_dict = {'bigrams': bi_cfd, 'trigrams': tri_cfd}
cache_dict['{}'.format(file)] = file_dict
cache_dict['{}'.format(file)].update(
{'hap': hapaxes, 'v_count': vocab_count})
cprint('====== {:.2f}s -- cached {}'.format(time.time() -
start_time, file), color='green')
cprint('\n\n=========== CACHE TOOKED {:.2f} seconds\n =========='.format(
time.time() - time_zero), color='yellow')
return cache_dict
def deteksi_ds(yolo):
# Definisikan Parameternya
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
tulis_video_output = True
# Model DEEP SORT diambil di sini..
namafile_model = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(namafile_model, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
video_capture = cv2.VideoCapture("demo-1.mp4")
video_capture.set(cv2.CAP_PROP_FRAME_WIDTH, 160)
video_capture.set(cv2.CAP_PROP_FRAME_HEIGHT, 120)
#
# Kalau ingin menulis hasil rekaman videonya buat VideoWriter object dgn codec-nya
# sekalian tulis juga hasil deteksi ke file txt
#
if tulis_video_output:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('hasil_output2.avi', fourcc, 15, (w, h))
list_file = open('hasil_deteksi.txt', 'w')
frame_index = -1
# --------------------
# MULAI CAPTURE VIDEO
# --------------------
# FPS awal
fps = 0.0
jum_track = set()
while True:
ret, frame = video_capture.read()
# Warna Gray
# frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
# baca frame
if ret != True:
break
# siapkan hitungan waktu
t1 = time.time()
# konversi bgr ke rgb
image = Image.fromarray(frame[..., ::-1])
kotak = yolo.detect_image(image)
# print("[-] Kotak : ",len(kotak))
features = encoder(frame, kotak)
# score to 1.0 here).
deteksi_box = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(kotak, features)
]
# panggil non-maxima suppression
boxes = np.array([d.tlwh for d in deteksi_box])
scores = np.array([d.confidence for d in deteksi_box])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
deteksi_box = [deteksi_box[i] for i in indices]
# Panggil class tracker
tracker.predict()
tracker.update(deteksi_box)
# Tracking hasil deteksi
for track in tracker.tracks:
# apakah berhasil di track?
if not track.is_confirmed() or track.time_since_update > 1:
continue
# buat bounding box-nya
bbox = track.to_tlbr()
# box
# Ini adalah prediction box
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
# teks untuk box_id
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 5e-3 * 200,
(0, 255, 0), 1)
jum_track.add(track.track_id)
cv2.putText(frame, "> Jumlah Orang : " + str(max(jum_track)),
(10, 25), cv2.FONT_HERSHEY_DUPLEX, .5, (0, 0, 255), 1)
print(">> Hits (Manusia) :", max(jum_track))
#
# pastikan box deteksi ada terus
# ini adalah Detection Box
for det in deteksi_box:
bbox = det.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
# tampilkan - DEBUG
cv2.imshow('Test Video', frame)
#
# Jika tulis video?
#
if tulis_video_output:
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index) + ' ')
if len(kotak) != 0:
for i in range(0, len(kotak)):
list_file.write(
str(kotak[i][0]) + ' ' + str(kotak[i][1]) + ' ' +
str(kotak[i][2]) + ' ' + str(kotak[i][3]) + ' ')
list_file.write('\n')
# tampilan fps
fps = (fps + (1. / (time.time() - t1))) / 2
print(">> fps= %f" % (fps))
# tekan Q untuk stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# bersihkan video_capture
video_capture.release()
if tulis_video_output:
# release video
out.release()
# tutup file
list_file.close()
# release semua
cv2.destroyAllWindows()
z = 1
x = f(0)
y = f(x)
for _ in xrange(iter1):
z = 1
for _ in xrange(iter2):
x = f(x) % n
y = f(f(y)) % n
z = (z * (x - y)) % n
d = gcd(z, n)
if d > 1:
return d, n//d
return d, n//d
# 187, 4717, 40100490452444053, 278009, 63053699, 549314599, 7247123747459, 2097335995683611, 4274010960572200553847767
start_time = time.time()
print Pollard(40100490452444053)
elapsed_time = time.time() - start_time
print("%0.10f" % elapsed_time)
def isqrt(n):
x = n
y = (x + 1) // 2
while y < x:
x = y
y = (x + n // x) // 2
return x
def Fermat(n):
"""
return frame
net = build_ssd('test', 300, 21) # initialize SSD
net.load_state_dict(torch.load('data/weights/ssd_300_VOC0712.pth'))
transform = BaseTransform(net.size, (104 / 256.0, 117 / 256.0, 123 / 256.0))
video_capture = cv2.VideoCapture(webcam_index)
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if frame.shape[0] == 0:
break
t1 = time.time()
# rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
# bboxes = face_detector.predict(rgb_frame, thresh)
ann_frame = predict(frame)
bg = np.zeros((map_height, map_width, 3))
# for box in bboxes:
# actual_face_size = math.sqrt(box[2] * box[3]) # use area of the face size as a measure of distance
# distance = max_face_size - actual_face_size # closest faces should be lowest values
# x = box[0] / max_x # scale to 0-1
# y = distance / max_y # scale to 0-1
# # print(str(x) + ", " + str(y))
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
width = 1280
height = 720
rfps = 10
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
tracking = args.tracking
writeVideo_flag = args.writeVideo_flag
asyncVideo_flag = args.asyncVideo_flag
webcamera_flag = args.webcamera_flag
ipcamera_flag = args.ipcamera_flag
udp_flag = args.udp_flag
full_cam_addr, key = sd.set_address(args.ipaddress, args.cam_ip,
args.cam_cmd, args.key)
cam_ip = full_cam_addr.replace(args.cam_cmd,
"").replace("rtsp://*****:*****@", "")
if args.jpegmode:
full_cam_addr = full_cam_addr.replace("rtsp", "http")
print(full_cam_addr)
print(key)
if asyncVideo_flag:
print("load videofile")
video_capture = VideoCaptureAsync(args.videofile)
elif ipcamera_flag or args.jpegmode:
print("load ipcamera")
video_capture = cv2.VideoCapture(full_cam_addr)
# width = video_capture.get(cv2.CAP_PROP_FRAME_WIDTH)
# height = video_capture.get(cv2.CAP_PROP_FRAME_HEIGHT)
# rfps = video_capture.get(cv2.CAP_PROP_FPS)
print("fps:{}width:{}height:{}".format(rfps, width, height))
elif webcamera_flag:
print("load webcamera")
video_capture = cv2.VideoCapture(0)
else:
print("load videofile")
video_capture = cv2.VideoCapture(args.videofile)
# video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
if udp_flag:
HOST = ''
PORT = 5000
address = '192.168.2.255'
sock = socket(AF_INET, SOCK_DGRAM)
sock.setsockopt(SOL_SOCKET, SO_BROADCAST, 1)
sock.bind((HOST, PORT))
fps = 0.0
i = 0
if not args.maskoff:
maskbgi = Image.new('RGB', (int(width), int(height)), (0, 0, 0))
mask = Image.open(args.maskdir + 'mask' + args.ipaddress[-1] +
'.png').convert("L").resize(size=(int(width),
int(height)),
resample=Image.NEAREST)
while True:
nowtime = datetime.datetime.now(
timezone('Asia/Tokyo')).strftime('%Y-%m-%d %H:%M:%S.%f%z')
ret, frame = video_capture.read() # frame shape 640*480*3
if not ret:
print('cant read')
video_capture = cv2.VideoCapture(full_cam_addr)
continue
t1 = time.time()
try:
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
except TypeError:
video_capture = cv2.VideoCapture(full_cam_addr)
continue
image = Image.composite(maskbgi, image, mask)
boxes, confidence, classes = yolo.detect_image(image)
if tracking:
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
else:
detections = [
Detection_YOLO(bbox, confidence, cls)
for bbox, confidence, cls in zip(boxes, confidence, classes)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
car_data = {}
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
car_data[str(track.track_id)] = [
int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3])
]
sd.send_amqp(
sd.create_jsondata(cam_ip, nowtime,
time.time() - t1, car_data, args.jsonfile,
args.json_path, i), key, args.AMQPHost)
i += 1
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
### 読み飛ばし処理を追加 ###
if not args.jsonfile and args.skip:
if fps <= 10:
for _i in range(int(math.ceil(rfps / fps)) - 1):
ret, frame = video_capture.read()
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo, sequence_file, fps_render_rate, writeVideo_flag, labels_file,
hide_window):
# Compute output file
file_name = os.path.splitext(
os.path.basename(sequence_file))[0] if sequence_file != '0' else '0'
if sequence_file == '0':
sequence_file = 0
# Compute the action map if labels provided
action_map = dict()
if labels_file != None:
action_map = parse_labels_file(labels_file)
print(action_map)
# Build directory path
frames_dir_path = "output/action_tubes/" + file_name
if not writeVideo_flag:
if os.path.exists(frames_dir_path):
shutil.rmtree(frames_dir_path)
os.mkdir(frames_dir_path)
# Create coords dir for movie
coords_path = 'output/tracked_bounding_boxes/' + file_name + '.json'
output_seq = 'output/annotated_videos/' + file_name + '.avi'
# Dict of coordinates for each tracked individual
track_map = dict()
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = 'object_detection/model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
video_capture = cv2.VideoCapture(sequence_file)
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID') #*'MJPG'
# Build video output handler only if we are not cropping
out = cv2.VideoWriter(output_seq, fourcc, fps_render_rate, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
fps = 0.0
frame_number = 0
while video_capture.isOpened():
frame_number += 1
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
# image = Image.fromarray(frame)
image = Image.fromarray(frame[..., ::-1]) #bgr to rgb
boxs = yolo.detect_image(image)
# print("box_num",len(boxs))
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
crop_img = frame[int(bbox[1]):int(bbox[3]),
int(bbox[0]):int(bbox[2])].copy()
# Append coordinates for individual to track map
if track.track_id not in track_map:
track_map[track.track_id] = [
(frame_number,
[int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3])])
]
else:
track_map[track.track_id].append(
(frame_number,
[int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3])]))
# Build directory path
frames_dir_path = "output/action_tubes/" + file_name + '/' + str(
track.track_id)
if not os.path.exists(frames_dir_path) and not writeVideo_flag:
os.mkdir(frames_dir_path)
# Write frame or annotate frame
if not writeVideo_flag:
cv2.imwrite(frames_dir_path + "/" + str(frame_number) + ".jpg",
crop_img)
else:
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
append_str = str(track.track_id) + ": Person"
if track.track_id in action_map:
append_str += ' ' + action_map[track.track_id]
cv2.putText(frame, append_str, (int(bbox[0]), int(bbox[1])), 0,
5e-3 * 200, (0, 255, 0), 2)
with open(coords_path, 'w') as fp:
json.dump(track_map, fp)
for det in detections:
bbox = det.to_tlbr()
if writeVideo_flag:
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
if not hide_window:
cv2.imshow('', cv2.resize(frame, (1200, 675)))
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index) + ' ')
if len(boxs) != 0:
for i in range(0, len(boxs)):
list_file.write(
str(boxs[i][0]) + ' ' + str(boxs[i][1]) + ' ' +
str(boxs[i][2]) + ' ' + str(boxs[i][3]) + ' ')
list_file.write('\n')
fps = (fps + (1. / (time.time() - t1))) / 2
print("fps= %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
tracking = True
writeVideo_flag = True
asyncVideo_flag = False
file_path = 'video.webm'
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
if tracking:
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
else:
detections = [
Detection_YOLO(bbox, confidence, cls)
for bbox, confidence, cls in zip(boxes, confidence, classes)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0,
1.5e-3 * frame.shape[0], (0, 255, 0), 1)
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2) + "%"
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
if len(classes) > 0:
cls = det.cls
cv2.putText(frame,
str(cls) + " " + score,
(int(bbox[0]), int(bbox[3])), 0,
1.5e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def detect_image(self, image):
start = time.time()
if self.is_fixed_size:
assert self.model_image_size[
0] % 32 == 0, 'Multiples of 32 required'
assert self.model_image_size[
1] % 32 == 0, 'Multiples of 32 required'
boxed_image = letterbox_image(
image, tuple(reversed(self.model_image_size)))
else:
new_image_size = (image.width - (image.width % 32),
image.height - (image.height % 32))
boxed_image = letterbox_image(image, new_image_size)
image_data = np.array(boxed_image, dtype='float32')
print(image_data.shape)
image_data /= 255.
image_data = np.expand_dims(image_data, 0) # Add batch dimension.
out_boxes, out_scores, out_classes = self.sess.run(
[self.boxes, self.scores, self.classes],
feed_dict={
self.yolo_model.input: image_data,
self.input_image_shape: [image.size[1], image.size[0]],
K.learning_phase(): 0
})
print('Found {} boxes for {}'.format(len(out_boxes), 'img'))
font = ImageFont.truetype(font='font/FiraMono-Medium.otf',
size=np.floor(3e-2 * image.size[1] +
0.5).astype('int32'))
thickness = (image.size[0] + image.size[1]) // 300
for i, c in reversed(list(enumerate(out_classes))):
predicted_class = self.class_names[c]
box = out_boxes[i]
score = out_scores[i]
label = '{} {:.2f}'.format(predicted_class, score)
draw = ImageDraw.Draw(image)
label_size = draw.textsize(label, font)
top, left, bottom, right = box
top = max(0, np.floor(top + 0.5).astype('int32'))
left = max(0, np.floor(left + 0.5).astype('int32'))
bottom = min(image.size[1], np.floor(bottom + 0.5).astype('int32'))
right = min(image.size[0], np.floor(right + 0.5).astype('int32'))
print(label, (left, top), (right, bottom))
if top - label_size[1] >= 0:
text_origin = np.array([left, top - label_size[1]])
else:
text_origin = np.array([left, top + 1])
# My kingdom for a good redistributable image drawing library.
for i in range(thickness):
draw.rectangle([left + i, top + i, right - i, bottom - i],
outline=self.colors[c])
draw.rectangle(
[tuple(text_origin),
tuple(text_origin + label_size)],
fill=self.colors[c])
draw.text(text_origin, label, fill=(0, 0, 0), font=font)
del draw
end = time.time()
print(end - start)
return image
def opt_aux(atom_list, jmats, spins, tol = 1.0e-10):
"""This function separates the functionality of the optimizer from the
files. This method assumes that jmats is in the correct order.
ie. jnums looks like [0,1,2,3...]"""
N_atoms = len(atom_list)
# Generate the Jij and anisotropy arrays
Jij = gen_Jij(atom_list,jmats)
anis = gen_anisotropy(atom_list)
# Get the spin magnitudes from the atoms in atom list
spin_mags = []
for atom in atom_list:
spin_mags.append(atom.spinMag)
spin_mags = np.array(spin_mags)
# hamiltonian method
def hamiltonian(p, Jij = None, spinMags = None, anis = None):
""" Computes the hamiltonian given a list a thetas and phis"""
# Thetas are the first half, phis the second half
theta = p[:len(p)//2]
phi = p[len(p)//2:]
# Sx,Sy,Sz
Sx = spinMags*sin(theta)*cos(phi)
Sy = spinMags*sin(theta)*sin(phi)
Sz = spinMags*cos(theta)
# print 'local opt spins'
# print Sx[0], Sy[0], Sz[0]
# Array of spin vectors for each atom. Reshape it. Calculate hamiltonian with it and return the hamiltonian.
Sij = np.array([Sx,Sy,Sz])
Sij = Sij.T.reshape(1,3*len(p)//2)[0].T
SijT = Sij.T
#res1 = SijT * Sij
res1 = SijT*Jij
Hij = np.dot(res1,Sij).flat[0]
Ham = - Hij - np.dot(anis, Sij**2)
return Ham
# derivative of the hamiltonian
def deriv(p, Jij = None, spinMags = None, anis = None):
""" Computes the derivative of the hamiltonian with respect to each theta and then each phi"""
# Thetas are the first half, phis the second half
half = len(p)/2
theta = p[:half]
phi = p[half:]
# Sx,Sy,Sz
Sx = spinMags*sin(theta)*cos(phi)
Sy = spinMags*sin(theta)*sin(phi)
Sz = spinMags*cos(theta)
# dSx/dtheta,dSy/dtheta,dSz/dtheta
Sxt = spinMags*cos(theta)*cos(phi)
Syt = spinMags*cos(theta)*sin(phi)
Szt = -spinMags*sin(theta)
# dSx/dphi,dSy/dphi,dSz/dphi
Sxp = -spinMags*sin(theta)*sin(phi)
Syp = spinMags*sin(theta)*cos(phi)
Szp = 0*cos(theta)
# Makes an array of the derivatives with respect to theta, then another with respect to phi
# (dSx1/dtheta1,dSy1/dtheta1,dSz1/dtheta1
# ( dSx2/dtheta2,dSy2/dtheta2,dSz2/dtheta2
# ( ...
# Similarly for phis
# Then we multiply this by Jij which yields a half by 3*half array. We then multiply this by
# the Sij vector which finally yields a half by 1 array. Thus
# dSijt * Jij * Sij -> [dE/dtheta1, dE/dtheta2, dE/dtheta3, ...]
# and similarly for phi
dSijt = np.zeros((half,3*half))
dSijp = np.zeros((half,3*half))
dSijt[range(half),range(0,3*half,3)]=Sxt
dSijt[range(half),range(1,3*half,3)]=Syt
dSijt[range(half),range(2,3*half,3)]=Szt
dSijp[range(half),range(0,3*half,3)]=Sxp
dSijp[range(half),range(1,3*half,3)]=Syp
dSijp[range(half),range(2,3*half,3)]=Szp
# Standard Sij spin vector we've been using
Sij = np.array([Sx,Sy,Sz])
Sij = Sij.T.reshape(1,3*len(p)//2)[0].T
# Calculate a hamiltonian for both theta and phi
res1t = np.dot(dSijt * Jij, Sij)
res2t = np.dot(Sij.T, Jij * dSijt.T)
Hijt = res1t + res2t
Hamt = - Hijt - np.matrix(np.dot(2*anis.T*Sij,dSijt.T))
Hamt = Hamt.T
res1p = np.dot(dSijp * Jij, Sij)
res2p = np.dot(Sij.T, Jij * dSijp.T)
Hijp = res1p + res2p
Hamp = - Hijp - np.matrix(np.dot(2*anis.T*Sij,dSijp.T))
Hamp = Hamp.T
# Concatenate the two and return the result
result = np.concatenate((np.array(Hamt),np.array(Hamp)))
return result.T
# populate initial p list
# returns a numpy array of all the thetas followed by all the phis
thetas = []
phis = []
if len(spins)!= N_atoms: raise Exception('poop')
for i in range(N_atoms):
sx = spins[i][0]
sy = spins[i][1]
sz = spins[i][2]
s = atom_list[i].spinMag
theta = arccos(sz/s)
phi = np.arctan2(sy,sx)
thetas.append(theta)
phis.append(phi)
# print 'initial spins'
# print s*sin(theta)*cos(phi), s*sin(theta)*sin(phi), s*cos(theta)
p0 = np.array(thetas+phis)
# define the limits parameter list
limits = []
for i in range(len(p0)):
if i < len(p0)//2:#theta
limits.append((0,pi))
else:#phi
limits.append((-pi,pi))
print "Local Optimization Beginning..."
print "tolerance", tol
st = time.time()
# call minimizing function
m = fmin_l_bfgs_b(hamiltonian, p0, fprime = deriv, args = (Jij, spin_mags, anis), pgtol=tol, bounds = limits)
print time.time()-st, "seconds"
print "Optimization Complete"
# grab returned parameters
# thetas are the first half of the list, phis are the second half
pout=m[0]
theta=pout[0:len(pout)/2]
phi=pout[len(pout)/2::]
# recreate sx,sy,sz's
sx=spin_mags*sin(theta)*cos(phi)
sy=spin_mags*sin(theta)*sin(phi)
sz=spin_mags*cos(theta)
return np.array([sx,sy,sz]).T
# Test if hysplit binary exists
hysplit_bin = 'C:\\hysplit4\\exec\\hyts_std.exe'
if not os.path.isfile(hysplit_bin):
print("Couldn't find hyts_std.exe. Please, check the hysplit installation or set\
properly the script variable 'hysplit_bin'.")
sys.exit()
print "Enter:"
meteo_dir = raw_input("Meteo directory (e.g. C:\\meteo): ")
output_dir = raw_input("Output directory (e.g. C:\\out): ")
csv_source = raw_input("Location of the csv file containing run specifications (e.g. C:\\runs.csv): ")
# Execution start time stamp
startTime = time.time()
# Load runs from csv file
csv_input = csv.reader(open(csv_source, 'r'))
# ASCDATA.CFG
ASCDATA = """-90.0 -180.0 lat/lon of lower left corner
1.0 1.0 lat/lon spacing in degrees
180 360 lat/lon number of data points
2 default land use category
0.2 default roughness length (m)
'C:/hysplit4/bdyfiles/' directory of files
"""
# SETUP.CFG
SETUP = """&SETUP\ntratio = 0.75,\nmgmin = 15,\nkhmax = 9999,\nkmixd = 0,
kmsl = 0,\nnstr = 0,\nmhrs = 9999,\nnver = 0,\ntout = 60,\ntm_tpot = 0,
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename,batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
video_capture = cv2.VideoCapture(0)
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('output.avi', fourcc, 15, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break;
t1 = time.time()
image = Image.fromarray(frame)
boxs = yolo.detect_image(image)
# print("box_num",len(boxs))
features = encoder(frame,boxs)
# score to 1.0 here).
detections = [Detection(bbox, 1.0, feature) for bbox, feature in zip(boxs, features)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if track.is_confirmed() and track.time_since_update >1 :
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(255,255,255), 2)
cv2.putText(frame, str(track.track_id),(int(bbox[0]), int(bbox[1])),0, 5e-3 * 200, (0,255,0),2)
for det in detections:
bbox = det.to_tlbr()
cv2.rectangle(frame,(int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(255,0,0), 2)
cv2.imshow('', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index)+' ')
if len(boxs) != 0:
for i in range(0,len(boxs)):
list_file.write(str(boxs[i][0]) + ' '+str(boxs[i][1]) + ' '+str(boxs[i][2]) + ' '+str(boxs[i][3]) + ' ')
list_file.write('\n')
fps = ( fps + (1./(time.time()-t1)) ) / 2
print("fps= %f"%(fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
def test():
global TOTAL_TIME
start = time.time()
im = mapnik.Image(m.width,m.height)
mapnik.render(m,im)
TOTAL_TIME += (time.time() - start)
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.2
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
video_capture = cv2.VideoCapture(0)
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame)
boxs = yolo.detect_image(image)
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track, det in zip(tracker.tracks, detections):
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 4)
bbox = det.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 4)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 5e-3 * 480,
(124, 252, 0), 4)
cv2.imshow('', frame)
fps = (fps + (1. / (time.time() - t1))) / 2
print("fps= %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
cv2.destroyAllWindows()
def prepare(self):
super(ComposeDebug, self).prepare()
self.timing_started = True
self.start_time = time.time()
self.debug_msg("Nik2img starting...")
self.debug_msg("Format: %s" % self.format)
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
writeVideo_flag = False
#video_capture = cv2.VideoCapture("rtsp://*****:*****@192.168.30.81:554/3")
#video_capture = cv2.VideoCapture("test.mp4")
video_capture = cv2.VideoCapture(0)
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('output.avi', fourcc, 15, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
fps = 0.0
htr = HumanTraceRecorder()
frameCounter = 0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
#frame = cv2.resize(frame,(640,360), interpolation=cv2.INTER_LINEAR)
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame)
boxs = yolo.detect_image(image)
# print("box_num",len(boxs))
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if track.is_confirmed() and track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
print(track.name_id)
if (frameCounter % 5) == 0:
x = int((bbox[0] + bbox[2]) / 2)
y = int((bbox[1] + bbox[3]) / 2)
new_time = frameCounter
if track.name_id != 0: # this is a tracked person
htr.updatePerson(track.name_id, x, y, new_time)
else:
subimage = image.crop((int(bbox[0]), int(bbox[1]),
int(bbox[2]), int(bbox[3])))
subimagearr = np.asarray(subimage)
faceID = getFaceID(
subimagearr,
predictor_path=
"./deep_sort/traceRecording/shape_predictor_5_face_landmarks.dat",
face_rec_model_path=
"./deep_sort/traceRecording/dlib_face_recognition_resnet_model_v1.dat"
)
if faceID == 0:
print("No face found")
else:
find, name_id = htr.checkPerson(faceID)
if find:
track.name_id = htr.updatePerson(
name_id, x, y, new_time)
else:
track.name_id = htr.addNewPerson(
faceID, x, y, new_time)
cv2.putText(frame, str(track.name_id),
(int(bbox[0]), int(bbox[1])), 0, 5e-3 * 200,
(0, 255, 0), 2)
frameCounter += 1
"""
if track.track_id != 0: # this is a tracked person
dataBase.addPersonTrace(track.track_id, location) # add the new location to the person's trace
else : # new detected person, need to verify the identity
subimage = image(cv2.Rect(int(bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3])))
result = faceDetect(subimage) # result should be the face data detected
if result: # make sure there is a face detection
if dataBase.checkPerson(result): # check if the person was recorded before
name = dataBase.checkName(result)
else: # new person, add to dataBase
newName = XXXXXXXXX # XXXXXXXXX is the new name
dataBase.addPerson(result, newName)
name = newName
dataBase.addPersonTrace(name, location) # add the new location to the person's trace
track.track_id = name
else : # no face detected, cannot identify the person
# do nothing, leave track unchange and wait for the next chance
"""
for det in detections:
bbox = det.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
cv2.imshow('', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index) + ' ')
if len(boxs) != 0:
for i in range(0, len(boxs)):
list_file.write(
str(boxs[i][0]) + ' ' + str(boxs[i][1]) + ' ' +
str(boxs[i][2]) + ' ' + str(boxs[i][3]) + ' ')
list_file.write('\n')
fps = (fps + (1. / (time.time() - t1))) / 2
print("fps= %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
htr.saveToFile()
def main(yolo):
points=[]
tpro=0.
# Definition of the parameters
max_cosine_distance = 0.9
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename,batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
video_capture = cv2.VideoCapture(0)
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('output.avi', fourcc, 15, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break;
frame=cv2.flip(frame,1)
image = Image.fromarray(frame)
# ___________________________________________________________________________DETECT WITH YOLO
t1 = time.time()
boxs = yolo.detect_image(image)
# print("box_num",len(boxs))
features = encoder(frame,boxs)
# score to 1.0 here).
detections = [Detection(bbox, 1.0, feature) for bbox, feature in zip(boxs, features)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# ___________________________________________________________________________DRAW DETECT BOX
to_move=[]
for det in detections:
bbox = det.to_tlbr()
cv2.rectangle(frame,(int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(0,0,255), 1)
temp=int(bbox[0]),int(bbox[1]),int(bbox[2]),int(bbox[3])
to_move.append(temp)
# now feed tracked box to move
# ___________________________________________________________________________MOVE
if to_move :
# Initial co-ordinates of the object to be tracked
# Create the tracker object
mover = [dlib.correlation_tracker() for _ in range(len(to_move))]
# Provide the tracker the initial position of the object
[mover[i].start_track(frame, dlib.rectangle(*rect)) for i, rect in enumerate(to_move)] ## FEED FIRST BOX HERE
for i in range (0,100): ##### START LOOP MOVER
ret, frame = video_capture.read() # tempo
full_frame_mover=[]
frame=cv2.flip(frame,1) #tempo
# Update the mover
for i in range(len(mover)):
#_____________FEED NEW IMAGE
mover[i].update(frame)
#_________________DRAW
rect = mover[i].get_position()
pt1 = (int(rect.left()), int(rect.top()))
pt2 = (int(rect.right()), int(rect.bottom()))
cv2.rectangle(frame, pt1, pt2, (255, 255, 255), 3)
full_frame_mover.append((pt1,pt2))
#print(full_frame_mover) # finish 1 frame
# ___________________________________________________________________________Call the tracker
tracker.predict()
tracker.update(detections)
# ___________________________________________________________________________DRAW TRACK RECTANGLE
for track in tracker.tracks:
if track.is_confirmed() and track.time_since_update >1 :
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(255,255,255), 2)
cv2.putText(frame, str(track.track_id),(int(bbox[0]), int(bbox[1])+30),0, 5e-3 * 200, (0,255,0),3)
dot=int(int(bbox[0])+((int(bbox[2])-int(bbox[0]))/2)),int(bbox[3]-10)
cv2.circle(frame,dot, 10, (0,0,255), -1)
cv2.imshow('', frame)
# Continue until the user presses ESC key
if cv2.waitKey(1) == 27:
break
# END LOOP MOVER
# ___________________________________________________________________________Call the tracker
tracker.predict()
tracker.update(full_frame_mover)
# ___________________________________________________________________________DRAW TRACK RECTANGLE
for track in tracker.tracks:
if track.is_confirmed() and track.time_since_update >1 :
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(255,255,255), 2)
cv2.putText(frame, str(track.track_id),(int(bbox[0]), int(bbox[1])+30),0, 5e-3 * 200, (0,255,0),3)
dot=int(int(bbox[0])+((int(bbox[2])-int(bbox[0]))/2)),int(bbox[3]-10)
cv2.circle(frame,dot, 10, (0,0,255), -1)
# ___________________________________________________________________________GET POINTS From click
if(cv2.waitKey(1)==ord('p')):
points = get_lines.run(frame, multi=True)
print(points)
if points :
for line in points:
cv2.line(frame, line[0:2], line[2:5], (0,255,255), 2) # draw line
cv2.imshow('', frame)
print('process time : ',time.time()-tpro)
tpro=time.time()
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index)+' ')
if len(boxs) != 0:
for i in range(0,len(boxs)):
list_file.write(str(boxs[i][0]) + ' '+str(boxs[i][1]) + ' '+str(boxs[i][2]) + ' '+str(boxs[i][3]) + ' ')
list_file.write('\n')
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
if not os.path.isfile(hysplit_bin):
print(
"Couldn't find hyts_std.exe. Please, check the hysplit installation or set\
properly the script variable 'hysplit_bin'.")
sys.exit()
print "Enter:"
meteo_dir = raw_input("Meteo directory (e.g. C:\\meteo): ")
output_dir = raw_input("Output directory (e.g. C:\\out): ")
csv_source = raw_input(
"Location of the csv file containing run specifications (e.g. C:\\runs.csv): "
)
# Execution start time stamp
startTime = time.time()
# Load runs from csv file
csv_input = csv.reader(open(csv_source, 'r'))
# ASCDATA.CFG
ASCDATA = """-90.0 -180.0 lat/lon of lower left corner
1.0 1.0 lat/lon spacing in degrees
180 360 lat/lon number of data points
2 default land use category
0.2 default roughness length (m)
'C:/hysplit4/bdyfiles/' directory of files
"""
# SETUP.CFG
SETUP = """&SETUP\ntratio = 0.75,\nmgmin = 15,\nkhmax = 9999,\nkmixd = 0,
kmsl = 0,\nnstr = 0,\nmhrs = 9999,\nnver = 0,\ntout = 60,\ntm_tpot = 0,
"DE"
DE = diffEvol(model, data);
# set_trace()
res = sorted([k for k in DE.DE()],
key = lambda F: F[-1])[-1]
return res
def tuner(model, data):
if model == rforest:
return _de(tuneRF, data)
elif model == CART:
return _de(tuneCART, data)
if __name__ == '__main__':
from timeit import time
data = explore(dir = '../Data/')[0][-1] # Only training data to tune.
for m in [tuneRF, tuneCART]:
t = time.time()
mdl = m(data)
# _test(data)
tunings = _de(m, data)
print tunings
print mdl.depen(tunings)
print time.time() - t
# print _de()
# print main()
# import sk; xtile = sk.xtile
# print xtile(G)
# main(dir = 'Data/')
'V':'$V_SW$ (km/s)', 'Vx':'$V_x$ (km/s)',
'Vy':'$V_y$ (km/s)', 'Vz':'$V_z$ (km/s)', # flow speed
'ProDen':'Proton Density ($N/cm^3$)',
'ProTmp':'Proton Temperature (k)',
'SYMD': 'SYM/D (nT)','SYMH': 'SYM/H (nT)',
'ASYD': 'ASY/D (nT)','ASYH': 'ASY/H (nT)',
'Kp':'Kp', 'R':'R','DST':'DST','ap':'ap','f107':'f10.7'}
omni_data = get_omni(bdate,edate,variables,res)
omni_data = pd.DataFrame(omni_data)
for k00, k0 in enumerate(variables):
plt.sca(ax[k00]) if len(variables)>1 else plt.sca(ax)
plt.plot(omni_data.index, omni_data[k0], **kwargs)
plt.ylabel(variablename[k0])
return
#END
#------------------------------------------------------------------------------
# TEST
if __name__ == '__main__':
# Test plot_omni
# fig,ax = plt.subplots(1,1,sharex=True)
# plot_omni(ax,'2010-1-1','2010-1-31',['Bx'],res='5minute')
# Test get_omni and get_omni2
from timeit import time
import matplotlib.pyplot as plt
begintime = time.time()
omni1 = get_omni('2010-1-1', '2010-12-31', ['AE'])
endtime = time.time()
print(endtime-begintime)
plt.plot(omni1['AE'])
def main(yolo):
start = time.time()
#Definition of the parameters
max_cosine_distance = 0.5 #余弦距离的控制阈值
nn_budget = None
nms_max_overlap = 0.3 #非极大抑制的阈值
counter = []
#deep_sort
model_filename = 'model_data/market1501.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
#video_path = "./output/output.avi"
video_capture = cv2.VideoCapture(args["input"])
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter(
'./output/' + args["input"][43:57] + "_" + args["class"] +
'_output.avi', fourcc, 15, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
# image = Image.fromarray(frame)
image = Image.fromarray(frame[..., ::-1]) #bgr to rgb
boxs, class_names = yolo.detect_image(image)
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
i = int(0)
indexIDs = []
c = []
boxes = []
for det in detections:
bbox = det.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
#boxes.append([track[0], track[1], track[2], track[3]])
indexIDs.append(int(track.track_id))
counter.append(int(track.track_id))
bbox = track.to_tlbr()
color = [int(c) for c in COLORS[indexIDs[i] % len(COLORS)]]
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (color), 3)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1] - 50)), 0, 5e-3 * 150,
(color), 2)
if len(class_names) > 0:
class_name = class_names[0]
cv2.putText(frame, str(class_names[0]),
(int(bbox[0]), int(bbox[1] - 20)), 0, 5e-3 * 150,
(color), 2)
i += 1
#bbox_center_point(x,y)
center = (int(
((bbox[0]) + (bbox[2])) / 2), int(((bbox[1]) + (bbox[3])) / 2))
#track_id[center]
pts[track.track_id].append(center)
thickness = 5
#center point
cv2.circle(frame, (center), 1, color, thickness)
#draw motion path
for j in range(1, len(pts[track.track_id])):
if pts[track.track_id][j - 1] is None or pts[
track.track_id][j] is None:
continue
thickness = int(np.sqrt(64 / float(j + 1)) * 2)
cv2.line(frame, (pts[track.track_id][j - 1]),
(pts[track.track_id][j]), (color), thickness)
#cv2.putText(frame, str(class_names[j]),(int(bbox[0]), int(bbox[1] -20)),0, 5e-3 * 150, (255,255,255),2)
count = len(set(counter))
cv2.putText(frame, "Total Object Counter: " + str(count),
(int(20), int(120)), 0, 5e-3 * 200, (0, 255, 0), 2)
cv2.putText(frame, "Current Object Counter: " + str(i),
(int(20), int(80)), 0, 5e-3 * 200, (0, 255, 0), 2)
cv2.putText(frame, "FPS: %f" % (fps), (int(20), int(40)), 0,
5e-3 * 200, (0, 255, 0), 3)
cv2.namedWindow("YOLO3_Deep_SORT", 0)
cv2.resizeWindow('YOLO3_Deep_SORT', 1024, 768)
cv2.imshow('YOLO3_Deep_SORT', frame)
if writeVideo_flag:
#save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index) + ' ')
if len(boxs) != 0:
for i in range(0, len(boxs)):
list_file.write(
str(boxs[i][0]) + ' ' + str(boxs[i][1]) + ' ' +
str(boxs[i][2]) + ' ' + str(boxs[i][3]) + ' ')
list_file.write('\n')
fps = (fps + (1. / (time.time() - t1))) / 2
#print(set(counter))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
print(" ")
print("[Finish]")
end = time.time()
if len(pts[track.track_id]) != None:
print(args["input"][43:57] + ": " + str(count) + " " +
str(class_name) + ' Found')
else:
print("[No Found]")
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
def main(yolo,read_type):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename,batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
#writeVideo_flag = True
#geneate a video object
video_dir='./model_data/demo2.wmv'
video=video_open(read_type,video_dir)
video_capture = video.generate_video()
fps=0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break;
t1 = time.time()
# 1、yolov3进行目标检测，找到目标位置和相应信息
# 2、跟踪物体在图像中的变化轨迹
# 1、yolov3进行目标检测，找到目标位置和相应信息
image = Image.fromarray(frame)
time3=time.time()
boxs = yolo.detect_image(image)
time4=time.time()
print('detect cost is',time4-time3)
# print("box_num",len(boxs))
time3=time.time()
features = encoder(frame,boxs)
# score to 1.0 here).
detections = [Detection(bbox, 1.0, feature) for bbox, feature in zip(boxs, features)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
time4=time.time()
print('features extract is',time4-time3)
# Call the tracker
# 2、跟踪物体在图像中的变化轨迹。利用卡尔曼滤波进行位置修正
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if track.is_confirmed() and track.time_since_update >1 :
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(255,255,255), 2)
cv2.putText(frame, str(track.track_id),(int(bbox[0]), int(bbox[1])),0, 5e-3 * 200, (0,255,0),2)
for det in detections:
bbox = det.to_tlbr()
cv2.rectangle(frame,(int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(255,0,0), 2)
cv2.imshow('', frame)
fps = ( fps + (1./(time.time()-t1)) ) / 2
print("fps= %f"%(fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
cv2.destroyAllWindows()
def main(yolo):
start = time.time()
max_cosine_distance = 0.5
nn_budget = None
nms_max_overlap = 0.3
counter = []
#deep_sort
model_filename = 'model_data/market1501.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
find_objects = ['person']
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
video_capture = cv2.VideoCapture(args["input"])
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('./output/output.avi', fourcc, 15, (w, h))
list_file = open('detection_rslt.txt', 'w')
frame_index = -1
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
classIDs = []
#image = Image.fromarray(frame)
image = Image.fromarray(frame[..., ::-1]) #bgr to rgb
boxs, confidence, class_names = yolo.detect_image(image)
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
i = int(0)
indexIDs = []
c = []
boxes = []
center2 = []
co_info = []
x_l = []
y_l = []
s_close_pair = []
for det in detections:
bbox = det.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
#print(class_names)
#print(class_names[p])
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
#boxes.append([track[0], track[1], track[2], track[3]])
indexIDs.append(int(track.track_id))
counter.append(int(track.track_id))
bbox = track.to_tlbr()
color = [int(c) for c in COLORS[indexIDs[i] % len(COLORS)]]
#print(frame_index)
list_file.write(str(frame_index) + ',')
list_file.write(str(track.track_id) + ',')
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (color), 3)
b0 = str(bbox[0]
) #.split('.')[0] + '.' + str(bbox[0]).split('.')[0][:1]
b1 = str(bbox[1]
) #.split('.')[0] + '.' + str(bbox[1]).split('.')[0][:1]
b2 = str(bbox[2] - bbox[0]
) #.split('.')[0] + '.' + str(bbox[3]).split('.')[0][:1]
b3 = str(bbox[3] - bbox[1])
list_file.write(
str(b0) + ',' + str(b1) + ',' + str(b2) + ',' + str(b3))
#print(str(track.track_id))
list_file.write('\n')
#list_file.write(str(track.track_id)+',')
cv2.putText(frame, "ID:" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 50)), 0, 5e-3 * 150,
(color), 2)
if len(class_names) > 0:
class_name = class_names[0]
cv2.putText(frame, str(class_names[0]),
(int(bbox[0]), int(bbox[1] - 20)), 0, 5e-3 * 150,
(color), 2)
i += 1
#bbox_center_point(x,y)
center = (int(
((bbox[0]) + (bbox[2])) / 2), int(((bbox[1]) + (bbox[3])) / 2))
#track_id[center]
pts[track.track_id].append(center)
thickness = 5
# draw distance line
(w, h) = (bbox[2], bbox[3])
center2.append(center)
co_info.append([w, h, center2])
#print(center2)
#calculateDistance
if len(center2) > 2:
for i in range(len(center2)):
for j in range(len(center2)):
#g = isclose(co_info[i],co_info[j])
#D = dist.euclidean((center2[i]), (center2[j]))
x1 = center2[i][0]
y1 = center2[i][1]
x2 = center2[j][0]
y2 = center2[j][1]
dis = calculateDistance(x1, y1, x2, y2)
if dis < 200:
#print(dis)
cv2.line(frame, (center2[i]), (center2[j]),
(0, 128, 255), 2)
if dis < 100:
#x_l.append(center2[i])
cv2.line(frame, (center2[i]), (center2[j]),
(0, 0, 255), 5)
#cv2.putText(frame, "KEEP DISTANCE",(int(960), int(1060)),0, 5e-3 * 200, (0,0,255),2)
else:
pass
#center point
cv2.circle(frame, (center), 1, color, thickness)
# draw motion path
for j in range(1, len(pts[track.track_id])):
if pts[track.track_id][j - 1] is None or pts[
track.track_id][j] is None:
continue
thickness = int(np.sqrt(64 / float(j + 1)) * 2)
#cv2.line(frame,(pts[track.track_id][j-1]), (pts[track.track_id][j]),(color),thickness)
count = len(set(counter))
cv2.putText(frame, "Total Pedestrian Counter: " + str(count),
(int(20), int(120)), 0, 5e-3 * 200, (0, 255, 0), 2)
cv2.putText(frame, "Current Pedestrian Counter: " + str(i),
(int(20), int(80)), 0, 5e-3 * 200, (0, 255, 0), 2)
cv2.putText(frame, "FPS: %f" % (fps * 2), (int(20), int(40)), 0,
5e-3 * 200, (0, 255, 0), 3)
cv2.namedWindow("YOLO3_Deep_SORT", 0)
cv2.resizeWindow('YOLO3_Deep_SORT', 1024, 768)
cv2.imshow('YOLO3_Deep_SORT', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps = (fps + (1. / (time.time() - t1))) / 2
out.write(frame)
frame_index = frame_index + 1
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
print(" ")
print("[Finish]")
end = time.time()
if len(pts[track.track_id]) != None:
print(args["input"][43:57] + ": " + str(count) + " " +
str(class_name) + ' Found')
else:
print("[No Found]")
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
video_capture = cv2.VideoCapture('top_view1.avi')
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('output1.avi', fourcc, 15, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
fps = 0.0
fig = plt.figure()
count = 0
x_list = []
y_list = []
# ax1 = fig.add_subplot(1, 1, 1)
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
print('NO VIDEO FOUND')
break
t1 = time.time()
# image = Image.fromarray(frame)
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxs = yolo.detect_image(image)
# print("box_co-ordinate = ", (boxs))
# for i in boxs:
# print(i[0][0])
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 5e-3 * 200,
(0, 255, 0), 2)
for det in detections:
bbox = det.to_tlbr()
# print((type(bbox)))
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
# print("The co-ordinates are:", int(bbox[0]), int(bbox[1]))
try:
for i in boxs:
x = (i[0] + i[2]) / 2
y = (i[1] + i[3]) / 2
count += 1
x_list.append(x)
y_list.append(y)
if count == 1:
points = plt.scatter(x_list, y_list)
elif count > 1:
print('x:', x_list, 'y:', y_list)
points.remove()
points = plt.scatter(x_list, y_list)
# plt.pause(0.9)
x_list.clear()
y_list.clear()
except:
continue
# redraw the canvas
fig.canvas.draw()
# convert canvas to image
img = np.fromstring(fig.canvas.tostring_rgb(), dtype=np.uint8, sep='')
img = img.reshape(fig.canvas.get_width_height()[::-1] + (3, ))
# img is rgb, convert to opencv's default bgr
img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
# display image with opencv or any operation you like
cv2.imshow("plot", img)
cv2.imshow('', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index) + ' ')
if len(boxs) != 0:
for i in range(0, len(boxs)):
list_file.write(
str(boxs[i][0]) + ' ' + str(boxs[i][1]) + ' ' +
str(boxs[i][2]) + ' ' + str(boxs[i][3]) + ' ')
list_file.write('\n')
fps = (fps + (1. / (time.time() - t1))) / 2
print("fps= %f" % (fps))
# Press Q to stop!
if cv2.waitKey(5) & 0xFF == ord('q'):
break
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
def dealVideoWithYoloAndDeepSort(yolo, video_list):
# Definition of the parameters
for video_info in video_list:
# 对应监控地点的阈值
address = video_info['address']
video_path = video_info['video_path']
addr = Address.objects.filter(
address=address).first() # video_info传过来是字符串，eval将其转化成字典
threshold = addr.threshold
VID = -1
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = 'edgeapp/model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric(
"cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
# 视频路径 若为0则检测摄像头
video_capture = cv2.VideoCapture(video_path)
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
out_video_fps = 30
# avi 格式
# fourcc = cv2.VideoWriter_fourcc(*'MJPG')
# out = cv2.VideoWriter('output.avi', fourcc, out_video_fps, (w, h))
# 生成mp4格式的输出视频 此“不支持”html5的 video标签播放
# fourcc = cv2.VideoWriter_fourcc(*'mp4v')
# out = cv2.VideoWriter('output.mp4', fourcc, out_video_fps, (w, h))
# 生成mp4格式的输出视频 此“支持”html5的 video标签播放
fourcc = cv2.VideoWriter_fourcc(*'avc1')
out = cv2.VideoWriter(
'edgeapp/Result_Video/{}_output.mp4'.format(address), fourcc,
out_video_fps, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
# image = Image.fromarray(frame)
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxs = yolo.detect_image(image)
# 统计人数
person_count = len(boxs)
print("视频中的人数：{}".format(person_count))
## HTTP 业务逻辑处理
if (person_count >= threshold) and (VID == -1):
# 让web服务器新建立一个异常视频对象
print("------开始建立异常视频对象")
res = requests.get("http://127.0.0.1:8080/push_new/",
params={
"vid": VID,
"address": address,
"number": person_count
})
res_dic = eval(res.text)
VID = res_dic['vid']
if VID != -1:
# 持续向web服务器推送当前监控地点的人数
print("------推送人数")
requests.get("http://127.0.0.1:8080/deal_new/",
params={
"vid": VID,
"address": address,
"number": person_count
})
features = encoder(frame, boxs)
# score to 1.0 here.
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(
boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 5e-3 * 200,
(0, 255, 0), 2)
for det in detections:
bbox = det.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
# 是否实时同步展示标记过的视频
# cv2.imshow('', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index) + ' ')
if len(boxs) != 0:
for i in range(0, len(boxs)):
list_file.write(
str(boxs[i][0]) + ' ' + str(boxs[i][1]) + ' ' +
str(boxs[i][2]) + ' ' + str(boxs[i][3]) + ' ')
list_file.write('\n')
fps = (fps + (1. / (time.time() - t1))) / 2
print("fps= %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
# 传送视频
send_video = open('edgeapp/Result_Video/{}_output.mp4'.format(address),
'rb')
files = {'file': send_video}
res = requests.post("http://127.0.0.1:8080/store_video/",
files=files,
data={"vid": VID})
# 关闭文件
send_video.close()
# 销毁yolo对象
del yolo
def main():
t0 = time.time()
problem = setup()
## Base Input Values
output = problem.objective()
print output
Plot_Mission.plot_mission(problem, -1)
# # Uncomment to view contours of the design space
# variable_sweep(problem)
print ' '
print ' Initial Guess Results '
print ' '
print 'Fuel Burn = ', float(problem.summary.base_mission_fuelburn)
print 'Cruise Fuel = ', float(problem.summary.cruise_fuel)
print 'Block Fuel = ', float(problem.summary.block_fuel)
print 'MTOW = ', float(problem.summary.MTOW)
print 'BOW = ', float(problem.summary.BOW)
print 'TOFL = ', float(problem.summary.takeoff_field_length)
print 'GRAD = ', float(problem.summary.second_segment_climb_gradient_takeoff)
print 'Cruise Alt = ', float(problem.summary.cruise_altitude)
print 'Design Ran = ', float(problem.summary.design_range)
print 'Cruise Ran = ', float(problem.summary.cruise_range)
print 'Total Ran = ', float(problem.summary.total_range)
print 'Time To Cli = ', float(problem.summary.time_to_climb_value)
print 'TOW HH = ', float(problem.summary.TOW_HH)
print 'Fuel HH = ', float(problem.summary.FUEL_HH)
print ' '
print 'Constraints = ', problem.all_constraints()
print ' '
last_inputs = problem.last_inputs[:, 1]
print 'S = ', last_inputs[0]
print 'AR = ', last_inputs[1]
print 't/c = ', last_inputs[2]
print 'Sweep Ang = ', last_inputs[3]
# ------------------------------------------------------------------
# Pareto
fuelburn = []
allconstraints = []
MTOW = []
finalvalue = []
grad = []
tofl = []
# betas = [1., 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.]
betas = [0.5, 0.25, 0.]
# betas = [0.]
fid = open('Pareto_results.txt', 'w') # Open output file
fid.write(' Pareto results \n')
fid.close()
for i, beta in enumerate(betas):
fid = open('Pareto_results.txt', 'ab') # Open output file
fid.write('Pareto Frontier. Run number: ' + str(i) + ' \n')
print('Pareto Frontier. Run number: ' + str(i))
# updating Beta value
design_vars = problem.optimization_problem.inputs[:, 1]
design_vars[-1] = beta
design_vars[0] = 75.
design_vars[1] = 8.0
design_vars[2] = 0.11
design_vars[3] = 20.
problem.optimization_problem.inputs[:, 1] = design_vars
bounds = problem.optimization_problem.inputs[:, 2]
bounds[-1] = list(bounds[-1])
bounds[-1][0] = beta
bounds[-1][1] = beta
bounds[-1] = tuple(bounds[-1])
problem.optimization_problem.inputs[:, 2] = bounds
output = scipy_setup.SciPy_Solve(problem, solver='SLSQP')
finalvalue.append(output)
print output
print ' '
print ' Final Results '
print ' '
print 'Fuel Burn = ', float(problem.summary.base_mission_fuelburn)
print 'Cruise Fuel = ', float(problem.summary.cruise_fuel)
print 'Block Fuel = ', float(problem.summary.block_fuel)
print 'MTOW = ', float(problem.summary.MTOW)
print 'BOW = ', float(problem.summary.BOW)
print 'TOFL = ', float(problem.summary.takeoff_field_length)
print 'GRAD = ', float(problem.summary.second_segment_climb_gradient_takeoff)
print 'Cruise Alt = ', float(problem.summary.cruise_altitude)
print 'Design Ran = ', float(problem.summary.design_range)
print 'Cruise Ran = ', float(problem.summary.cruise_range)
print 'Total Ran = ', float(problem.summary.total_range)
print 'Time To Cli = ', float(problem.summary.time_to_climb_value)
print 'TOW HH = ', float(problem.summary.TOW_HH)
print 'Fuel HH = ', float(problem.summary.FUEL_HH)
print ' '
print 'Constraints = ', problem.all_constraints()
print ' '
last_inputs = finalvalue[-1]
print 'S = ', last_inputs[0]
print 'AR = ', last_inputs[1]
print 't/c = ', last_inputs[2]
print 'Sweep Ang = ', last_inputs[3]
# -----------------------------------------
Plot_Mission.plot_mission(problem, i)
fuelburn.append(problem.summary.base_mission_fuelburn)
allconstraints.append(problem.all_constraints())
grad.append(problem.summary.second_segment_climb_gradient_takeoff)
tofl.append(problem.summary.takeoff_field_length)
MTOW.append(problem.summary.MTOW)
fid.write(str(fuelburn[-1])+' \n')
fid.write(str(grad[-1]) + ' \n')
fid.write(str(tofl[-1]) + ' \n')
fid.write(str(MTOW[-1]) + ' \n')
fid.write(str(allconstraints[-1]) + ' \n')
fid.write(str(finalvalue[-1]) + ' \n')
fid.write('\n \n')
fid.close()
fid = open('Pareto_results.txt', 'ab') # Open output file
elapsed = time.time() - t0
fid.write('Total run time: ' + str(elapsed))
print('Total run time: ' + str(elapsed))
fid.close()
return
def main(yolo):
print('Using {} model'.format(yolo))
# Definition of the parameters
max_cosine_distance = 0.2
nn_budget = None
nms_max_overlap = 0.4
# deep_sort
model_filename = 'model_data/models/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename,batch_size=1) # use to get feature
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric, max_age=100)
output_frames = []
output_rectanger = []
output_areas = []
output_wh_ratio = []
is_vis = True
out_dir = 'videos/output/'
print('The output folder is',out_dir)
if not os.path.exists(out_dir):
os.mkdir(out_dir)
all_frames = []
for video in args.videos:
loadvideo = LoadVideo(video)
video_capture, frame_rate, w, h = loadvideo.get_VideoLabels()
while True:
ret, frame = video_capture.read()
if ret != True:
video_capture.release()
break
all_frames.append(frame)
frame_nums = len(all_frames)
tracking_path = out_dir+'tracking'+'.avi'
combined_path = out_dir+'allVideos'+'.avi'
if is_vis:
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter(tracking_path, fourcc, frame_rate, (w, h))
out2 = cv2.VideoWriter(combined_path, fourcc, frame_rate, (w, h))
#Combine all videos
for frame in all_frames:
out2.write(frame)
out2.release()
#Initialize tracking file
filename = out_dir+'/tracking.txt'
open(filename, 'w')
fps = 0.0
frame_cnt = 0
t1 = time.time()
track_cnt = dict()
images_by_id = dict()
ids_per_frame = []
for frame in all_frames:
image = Image.fromarray(frame[...,::-1]) #bgr to rgb
boxs = yolo.detect_image(image) # n * [topleft_x, topleft_y, w, h]
features = encoder(frame,boxs) # n * 128
detections = [Detection(bbox, 1.0, feature) for bbox, feature in zip(boxs, features)] # length = n
text_scale, text_thickness, line_thickness = get_FrameLabels(frame)
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.delete_overlap_box(boxes, nms_max_overlap, scores) #preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices] # length = len(indices)
# Call the tracker
tracker.predict()
tracker.update(detections)
tmp_ids = []
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
area = (int(bbox[2]) - int(bbox[0])) * (int(bbox[3]) - int(bbox[1]))
if bbox[0] >= 0 and bbox[1] >= 0 and bbox[3] < h and bbox[2] < w:
tmp_ids.append(track.track_id)
if track.track_id not in track_cnt:
track_cnt[track.track_id] = [[frame_cnt, int(bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]), area]]
images_by_id[track.track_id] = [frame[int(bbox[1]):int(bbox[3]), int(bbox[0]):int(bbox[2])]]
else:
track_cnt[track.track_id].append([frame_cnt, int(bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]), area])
images_by_id[track.track_id].append(frame[int(bbox[1]):int(bbox[3]), int(bbox[0]):int(bbox[2])])
cv2_addBox(track.track_id,frame,int(bbox[0]),int(bbox[1]),int(bbox[2]),int(bbox[3]),line_thickness,text_thickness,text_scale)
write_results(filename,'mot',frame_cnt+1,str(track.track_id),int(bbox[0]),int(bbox[1]),int(bbox[2]),int(bbox[3]),w,h)
ids_per_frame.append(set(tmp_ids))
# save a frame
if is_vis:
out.write(frame)
t2 = time.time()
frame_cnt += 1
print(frame_cnt, '/', frame_nums)
if is_vis:
out.release()
print('Tracking finished in {} seconds'.format(int(time.time() - t1)))
print('Tracked video : {}'.format(tracking_path))
print('Combined video : {}'.format(combined_path))
os.environ["CUDA_VISIBLE_DEVICES"] = "0,1,2,3"
reid = REID()
threshold = 320
exist_ids = set()
final_fuse_id = dict()
print('Total IDs = ',len(images_by_id))
feats = dict()
for i in images_by_id:
print('ID number {} -> Number of frames {}'.format(i, len(images_by_id[i])))
feats[i] = reid._features(images_by_id[i]) #reid._features(images_by_id[i][:min(len(images_by_id[i]),100)])
ids_per_frame2 = copy.deepcopy(ids_per_frame)
for f in ids_per_frame:
if f:
if len(exist_ids) == 0:
for i in f:
final_fuse_id[i] = [i]
exist_ids = exist_ids or f
else:
new_ids = f-exist_ids
for nid in new_ids:
dis = []
if len(images_by_id[nid])<10:
exist_ids.add(nid)
continue
unpickable = []
for i in f:
for key,item in final_fuse_id.items():
if i in item:
unpickable += final_fuse_id[key]
print('exist_ids {} unpickable {}'.format(exist_ids,unpickable))
for oid in (exist_ids-set(unpickable))&set(final_fuse_id.keys()):
tmp = np.mean(reid.compute_distance(feats[nid],feats[oid]))
print('nid {}, oid {}, tmp {}'.format(nid, oid, tmp))
dis.append([oid, tmp])
exist_ids.add(nid)
if not dis:
final_fuse_id[nid] = [nid]
continue
dis.sort(key=operator.itemgetter(1))
if dis[0][1] < threshold:
combined_id = dis[0][0]
images_by_id[combined_id] += images_by_id[nid]
final_fuse_id[combined_id].append(nid)
else:
final_fuse_id[nid] = [nid]
print('Final ids and their sub-ids:',final_fuse_id)
print('MOT took {} seconds'.format(int(time.time() - t1)))
t2 = time.time()
# To generate MOT for each person, declare 'is_vis' to True
is_vis=False
if is_vis:
print('Writing videos for each ID...')
output_dir = 'videos/output/tracklets/'
if not os.path.exists(output_dir):
os.mkdir(output_dir)
loadvideo = LoadVideo(combined_path)
video_capture,frame_rate, w, h = loadvideo.get_VideoLabels()
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
for idx in final_fuse_id:
tracking_path = os.path.join(output_dir, str(idx)+'.avi')
out = cv2.VideoWriter(tracking_path, fourcc, frame_rate, (w, h))
for i in final_fuse_id[idx]:
for f in track_cnt[i]:
video_capture.set(cv2.CAP_PROP_POS_FRAMES, f[0])
_, frame = video_capture.read()
text_scale, text_thickness, line_thickness = get_FrameLabels(frame)
cv2_addBox(idx, frame, f[1], f[2], f[3], f[4], line_thickness, text_thickness, text_scale)
out.write(frame)
out.release()
video_capture.release()
# Generate a single video with complete MOT/ReID
if args.all:
loadvideo = LoadVideo(combined_path)
video_capture, frame_rate, w, h = loadvideo.get_VideoLabels()
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
complete_path = out_dir+'/Complete'+'.avi'
out = cv2.VideoWriter(complete_path, fourcc, frame_rate, (w, h))
for frame in range(len(all_frames)):
frame2 = all_frames[frame]
video_capture.set(cv2.CAP_PROP_POS_FRAMES, frame)
_, frame2 = video_capture.read()
for idx in final_fuse_id:
for i in final_fuse_id[idx]:
for f in track_cnt[i]:
#print('frame {} f0 {}'.format(frame,f[0]))
if frame == f[0]:
text_scale, text_thickness, line_thickness = get_FrameLabels(frame2)
cv2_addBox(idx, frame2, f[1], f[2], f[3], f[4], line_thickness, text_thickness, text_scale)
out.write(frame2)
out.release()
video_capture.release()
os.remove(combined_path)
print('\nWriting videos took {} seconds'.format(int(time.time() - t2)))
print('Final video at {}'.format(complete_path))
print('Total: {} seconds'.format(int(time.time() - t1)))
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
tracking = True
writeVideo_flag = False
asyncVideo_flag = False
webcamera_flag = True
ipcamera_flag = False
udp_flag = True
file_path = '/workspace/data/C0133_v4.mp4'
if asyncVideo_flag :
video_capture = VideoCaptureAsync(file_path)
elif ipcamera_flag :
video_capture = cv2.VideoCapture('rtsp://*****:*****@192.168.2.201/ONVIF/MediaInput?profile=def_profile1')
elif webcamera_flag :
video_capture = cv2.VideoCapture(0)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
if udp_flag:
HOST = ''
PORT = 5000
address = '192.168.2.255'
sock =socket(AF_INET, SOCK_DGRAM)
sock.setsockopt(SOL_SOCKET, SO_BROADCAST, 1)
sock.bind((HOST, PORT))
fps = 0.0
fps_imutils = imutils.video.FPS().start()
savetime = 0
while True:
nowtime = datetime.datetime.now().isoformat()
ret, frame = video_capture.read() # frame shape 640*480*3
t1 = time.time()
if time.time() - savetime >= 30:
print('save data')
cv2.imwrite("/workspace/images/image.png", frame)
savetime = time.time()
image = Image.fromarray(frame[...,::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
if tracking:
features = encoder(frame, boxes)
detections = [Detection(bbox, confidence, cls, feature) for bbox, confidence, cls, feature in
zip(boxes, confidence, classes, features)]
else:
detections = [Detection_YOLO(bbox, confidence, cls) for bbox, confidence, cls in
zip(boxes, confidence, classes)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id), (int(bbox[0]), int(bbox[1])), 0,
1.5e-3 * frame.shape[0], (0, 255, 0), 1)
# socket
message = str(nowtime + "," + str(track.track_id) + "," + str(int(bbox[0])) + "," + str(int(bbox[1])) + "," + str(int(bbox[2])) + "," + str(int(bbox[3])))
bmessage = message.encode('utf-8')
print(type(bmessage))
if udp_flag:
sock.sendto(message.encode('utf-8'), (address, PORT))
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2) + "%"
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
if len(classes) > 0:
cls = det.cls
cv2.putText(frame, str(cls) + " " + score, (int(bbox[0]), int(bbox[3])), 0,
1.5e-3 * frame.shape[0], (0, 255, 0), 1)
#cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1./(time.time()-t1))) / 2
print("FPS = %f"%(fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
class dataType:
Dixon = 'Dixon'
Rao = 'Rao'
CELLTYPE = cellType.K526
DataType = dataType.Rao
if CELLTYPE == 'hIMR90' or CELLTYPE == 'hES' or CELLTYPE == 'K526':
NUM_OF_CHRMS = 22
elif CELLTYPE == 'mES' or CELLTYPE == 'mCO':
NUM_OF_CHRMS = 20
for CHRM in range(1, NUM_OF_CHRMS + 1):
if DataType == dataType.Rao:
print('Loading Chromosome ' + str(CHRM))
st = time.time()
if PANDAS_INSTALLED:
if RESOLUTION < 1000000:
name = str(int(RESOLUTION / 1000)) + 'kb'
else:
name = str(int(RESOLUTION / 1000000)) + 'mb'
chr1Data = pd.read_csv(
os.path.abspath(os.sep) + 'Dataset/' + CELLTYPE + '/' + name +
'_resolution_intrachromosomal/chr' + str(CHRM) +
'/MAPQGE30/chr' + str(CHRM) + '_' + name + '.RAWobserved',
sep='\t',
header=None)
chr1Data = chr1Data.values
chr1Data[:, 0:2] = np.floor(chr1Data[:, 0:2] / RESOLUTION)
knorm = pd.read_csv(
os.path.abspath(os.sep) + 'Dataset/' + CELLTYPE + '/' + name +
from random import randrange
from selection_sort import selection_sort
from bubble_sort import bubble_sort
from timeit import time
# Creo la lista
# Lista di partenza da ordinare
mylist = [randrange(1000) for _ in range(10000)]
# Calcolo i tempi del selection sort....
selection_sort_time = selection_sort(mylist.copy())
# Calcolo i tempi del bubble sort....
bubble_sort_time = bubble_sort(mylist.copy())
# Calcolo i tempi del sort di default di python....
newlist = mylist.copy()
start_time = time.time()
newlist.sort()
stop_time = time.time()
default_sort_time = stop_time - start_time
# Salvo i risultati su file
result_file = open("results.csv", "a")
result_file.write(
f"{len(mylist)},{selection_sort_time},{bubble_sort_time},{default_sort_time}\n"
)
result_file.close()
def findTADs(intgMAT):
st = time.time()
intgMAT2 = np.lib.pad(intgMAT, ((maxL, maxL), (maxL, maxL)),
'constant',
constant_values=0)
# find potential starts/ends of TADs
det_scores = np.zeros([N, 3], dtype=np.float64)
det_scores = det_score(intgMAT2, np.arange(N), [W, W, W])
S = np.sum(det_scores[:, 0]) / N
det_scores[:, 1] = det_scores[:, 1] / S
det_scores[:, 2] = det_scores[:, 2] / S
start_peaks = np.asarray(detect_peaks(det_scores[:, 2], mpd=2),
dtype=np.int64) #np.median(det_scores[:,1])
end_peaks = np.asarray(detect_peaks(det_scores[:, 1], mpd=2),
dtype=np.int64) #np.median(det_scores[:,2])
# we need to know the preceding end for each start
prev_end = np.zeros(len(start_peaks), dtype=np.int64)
k_old = 0
for i in range(len(start_peaks)):
k = k_old
for j in range(k, len(end_peaks)):
if end_peaks[k] <= start_peaks[i]:
k = k + 1
else:
break
if k == 0:
prev_end[i] = -1
k_old = 0
elif end_peaks[k - 1] < start_peaks[0]:
prev_end[i] = -1
k_old = k - 1
else:
prev_end[i] = k - 1
k_old = k - 1
# we need to know the preceding start for each end
prev_start = np.zeros(len(end_peaks), dtype=np.int64)
k_old = 0
for i in range(len(end_peaks)):
k = k_old
for j in range(k, len(start_peaks)):
if start_peaks[k] <= end_peaks[i]:
k = k + 1
else:
break
if k == 0:
prev_start[i] = -1
k_old = 0
else:
prev_start[i] = k - 1
k_old = k - 1
# pre-compute scores for start/end combinations
M1 = len(start_peaks)
M2 = len(end_peaks)
rep = []
indx = []
for i in range(M1):
ind = prev_end[i]
indx.append(min(ind + 1, len(end_peaks)))
if ind == -1:
rep.append(M2)
elif ind == M2:
rep.append(0)
else:
rep.append(M2 - ind - 1)
long_i = np.repeat(start_peaks, rep)
long_j = np.zeros(np.sum(rep), dtype=np.int64)
for i in range(M1):
start = sum(rep[0:i])
long_j[start:start + rep[i]] = end_peaks[indx[i]:len(end_peaks)]
long_l = long_j - long_i + 1
temp_s = score(
intgMAT2, long_i, long_l
) #long_i is the center of a domain; long_l is the length of that domain
# build a dense representation of the 'scores' but much smaller than N*N (for higher resolutions)
i2I = np.zeros(N, dtype=np.int64)
for i in range(len(start_peaks)):
i2I[start_peaks[i]] = i
j2J = np.zeros(N, dtype=np.int64)
for j in range(len(end_peaks)):
j2J[end_peaks[j]] = j
scores = np.zeros([len(start_peaks), len(end_peaks)], dtype=np.float64)
scores[i2I[long_i], j2J[long_j]] = temp_s
# Dynamic Programming applies on potential start/end TAD boundaries we have found above
T = np.zeros(len(end_peaks), dtype=np.float64)
backT = np.zeros(
len(end_peaks), dtype=np.int64
) # shows the index of the start point (out of all start points) for each end point
k_old = 0
for j in range(len(end_peaks)):
if prev_start[j] == -1:
backT[j] = -1
continue
k = k_old
kk = np.arange(int(prev_start[j]) + 1)
vals = parDP(scores, i2I, j2J, T, start_peaks, end_peaks, prev_end,
j, kk)
T[j] = np.max([np.max(vals), T[j - 1]])
backT[j] = np.argmax(vals)
k_old = k
# Extracting TADs based on backT
counter = len(end_peaks) - 1
tadCount = 0
TADx1 = []
TADx2 = []
while True: # backT[counter] != 0 and backT[counter] != -1:
if backT[counter] == -1:
break
elif backT[counter] == 0:
TADx2.append(end_peaks[counter])
TADx1.append(start_peaks[0])
tadCount = tadCount + 1
break
elif counter > 0:
TADx2.append(end_peaks[counter])
TADx1.append(start_peaks[backT[counter]])
tadCount = tadCount + 1
counter = prev_end[backT[counter]]
if prev_end[backT[counter]] == counter:
counter = counter - 1
if counter == -1:
break
elif counter == 0:
TADx2.append(end_peaks[counter])
TADx1.append(start_peaks[backT[counter]])
tadCount = tadCount + 1
break
print('Number of TADs:', tadCount)
TADx1.sort()
TADx2.sort()
TADx1 = (np.asarray(TADx1)).reshape([len(TADx1), 1])
TADx2 = (np.asarray(TADx2)).reshape([len(TADx2), 1])
TAD = np.concatenate((TADx1, TADx2), axis=1)
np.savetxt(CELLTYPE + '_nij_chr' + str(CHRM) + '_' + str(N),
TAD,
delimiter=' ',
fmt='%d')
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.2
nn_budget = None
nms_max_overlap = 0.4
# deep_sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename,
batch_size=1) # use to get feature
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric, max_age=10)
output_frames = []
output_rectanger = []
output_areas = []
output_wh_ratio = []
is_vis = True
videos = ['videos/0701/1_1.mp4']
all_frames = []
for video in videos:
video_capture = cv2.VideoCapture(video)
w = int(video_capture.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(video_capture.get(cv2.CAP_PROP_FRAME_HEIGHT))
frame_rate = video_capture.get(cv2.CAP_PROP_FPS)
while True:
ret, frame = video_capture.read()
if ret != True:
video_capture.release()
break
all_frames.append(frame)
frame_nums = len(all_frames)
if is_vis:
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
output_path = video.split('.')[0] + '_output' + '.avi'
out = cv2.VideoWriter(output_path, fourcc, frame_rate, (w, h))
fps = 0.0
frame_cnt = 0
t1 = time.time()
track_cnt = dict()
images_by_id = dict()
ids_per_frame = []
for frame in all_frames:
image = Image.fromarray(frame[..., ::-1]) #bgr to rgb
boxs = yolo.detect_image(image) # n * [topleft_x, topleft_y, w, h]
features = encoder(frame, boxs) # n * 128
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
] # length = n
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
#indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
indices = preprocessing.delete_overlap_box(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices] # length = len(indices)
# Call the tracker
tracker.predict()
tracker.update(detections)
tmp_ids = []
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
area = (int(bbox[2]) - int(bbox[0])) * (int(bbox[3]) -
int(bbox[1]))
if bbox[0] >= 0 and bbox[1] >= 0 and bbox[3] < h and bbox[2] < w:
tmp_ids.append(track.track_id)
if track.track_id not in track_cnt:
track_cnt[track.track_id] = [[
frame_cnt,
int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3]), area
]]
images_by_id[track.track_id] = [
frame[int(bbox[1]):int(bbox[3]),
int(bbox[0]):int(bbox[2])]
]
else:
track_cnt[track.track_id].append([
frame_cnt,
int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3]), area
])
images_by_id[track.track_id].append(
frame[int(bbox[1]):int(bbox[3]),
int(bbox[0]):int(bbox[2])])
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 2, (0, 255, 0), 2)
ids_per_frame.append(set(tmp_ids))
if is_vis:
# save a frame
out.write(frame)
t2 = time.time()
frame_cnt += 1
print(frame_cnt, '/', frame_nums)
if is_vis:
out.release()
for i in images_by_id:
print(i, len(images_by_id[i]))
os.environ["CUDA_VISIBLE_DEVICES"] = "3"
reid = REID()
threshold = 250
exist_ids = set()
final_fuse_id = dict()
for f in ids_per_frame:
if f:
if len(exist_ids) == 0:
for i in f:
final_fuse_id[i] = [i]
exist_ids = exist_ids or f
else:
new_ids = f - exist_ids
for nid in new_ids:
dis = []
if len(images_by_id[nid]) < 10:
exist_ids.add(nid)
continue
unpickable = []
for i in f:
for key, item in final_fuse_id.items():
if i in item:
unpickable += final_fuse_id[key]
for oid in (exist_ids - set(unpickable)) & set(
final_fuse_id.keys()):
tmp = np.mean(
reid.get_features(images_by_id[nid],
images_by_id[oid]))
print(nid, oid, tmp)
dis.append([oid, tmp])
exist_ids.add(nid)
if not dis:
final_fuse_id[nid] = [nid]
continue
dis.sort(key=operator.itemgetter(1))
if dis[0][1] < threshold:
combined_id = dis[0][0]
images_by_id[combined_id] += images_by_id[nid]
final_fuse_id[combined_id].append(nid)
else:
final_fuse_id[nid] = [nid]
print(final_fuse_id)
if is_vis:
print('writing video...')
output_dir = 'videos/' + videos[0].split('/')[-1].split('.')[0]
print(output_dir)
if not os.path.exists(output_dir):
os.mkdir(output_dir)
video_capture = cv2.VideoCapture(videos[0])
w = int(video_capture.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(video_capture.get(cv2.CAP_PROP_FRAME_HEIGHT))
frame_rate = video_capture.get(cv2.CAP_PROP_FPS)
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
for idx in final_fuse_id:
output_path = os.path.join(output_dir, str(idx) + '.avi')
out = cv2.VideoWriter(output_path, fourcc, frame_rate, (w, h))
for i in final_fuse_id[idx]:
print(idx, i)
for f in track_cnt[i]:
video_capture.set(cv2.CAP_PROP_POS_FRAMES, f[0])
_, frame = video_capture.read()
cv2.rectangle(frame, (f[1], f[2]), (f[3], f[4]),
(255, 0, 0), 2)
out.write(frame)
out.release()
video_capture.release()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 0.7
# deep_sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
webcam_flag = False
resize_flag = True
resize_size = (800, 450)
# some links from earthcam https://github.com/Crazycook/Working/blob/master/Webcams.txt https://www.vlcm3u.com/web-cam-live/
# video_url = 'https://videos3.earthcam.com/fecnetwork/lacitytours1.flv/chunklist_w683585821.m3u8' # HOLLYWOOD
# video_url = 'https://videos3.earthcam.com/fecnetwork/9974.flv/chunklist_w1421640637.m3u8' # NYC
# video_url = 'https://videos3.earthcam.com/fecnetwork/5775.flv/chunklist_w1803081483.m3u8' # NYC 2
# video_url = 'http://181.1.29.189:60001/cgi-bin/snapshot.cgi?chn=0&u=admin'
# video_url = 'https://videos-3.earthcam.com/fecnetwork/15559.flv/chunklist_w573709200.m3u8' # NYC 3
video_url = 'https://hddn01.skylinewebcams.com/live.m3u8?a=97psdt8nv2hsmclta3nuu4di94'
if webcam_flag:
video_capture = cv2.VideoCapture(0)
else:
video_capture = cv2.VideoCapture()
video_capture.set(cv2.CAP_PROP_BUFFERSIZE, 2)
video_capture.open(video_url)
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('output.avi', fourcc, 15, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
if resize_flag:
frame = cv2.resize(frame,
resize_size,
interpolation=cv2.INTER_AREA)
# image = Image.fromarray(frame)
image = Image.fromarray(frame[..., ::-1]) #bgr to rgb
boxs = yolo.detect_image(image)
# print("box_num",len(boxs))
if np.array(boxs).size > 0:
features = encoder(frame, np.array(boxs)[:, 0:4].tolist())
class_names = yolo.class_names
# score to 1.0 here).
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(
boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
#Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1]) - 10), 0, 5e-3 * 100,
(0, 0, 255), 2)
for det in detections:
bbox = det.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
cv2.putText(
frame, class_names[int(det.label)] + "(" +
str(round(det.score, 2)) + ")",
(int(bbox[0]), int(bbox[3])), 0, 5e-3 * 90, (255, 0, 0), 2)
#cv2.putText(frame, str(int(bbox[0])) + "-" + str(int(bbox[3])) ,(int(bbox[0]), int(bbox[3])),0, 5e-3 * 90, (0,0,255),2)
cv2.imshow('', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index) + ' ')
if len(boxs) != 0:
for i in range(0, len(boxs)):
list_file.write(
str(boxs[i][0]) + ' ' + str(boxs[i][1]) + ' ' +
str(boxs[i][2]) + ' ' + str(boxs[i][3]) + ' ')
list_file.write('\n')
fps = (fps + (1. / (time.time() - t1))) / 2
print("fps= %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# deep_sort
model_filename = './models/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
video_capture = cv2.VideoCapture(args.input_video)
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('/media/rudy/C0E28D29E28D252E/2017_taroko/终点前/video_YDXJ0101.avi', fourcc, 15, (w, h))
list_file = open('/media/rudy/C0E28D29E28D252E/2017_taroko/终点前/detection_YDXJ0101.txt', 'w')
frame_index = -1
fps = 0.0
n = 0
id_frame = 0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
if int(args.skip_frame) != n:
n += 1
continue
n = 0
t1 = time.time()
image = Image.fromarray(frame)
boxs, classes = yolo.detect_image(image)
for idb, box in enumerate(boxs):
cv2.rectangle(frame, (int(box[0]), int(box[1])), (int(box[0])+int(box[2]), int(box[1])+int(box[3])),(255,255,255), 2)
cv2.putText(frame, str(classes[idb]),(int(box[0]), int(box[1])),0, 5e-3 * 100, (0,255,0),2)
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [Detection(bbox, 1.0, feature) for bbox, feature in zip(boxs, features)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if track.is_confirmed() and track.time_since_update > 1:
continue
bbox = track.to_tlbr()
# cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
# cv2.putText(frame, str(track.track_id), (int(bbox[0]), int(bbox[1])), 0, 5e-3 * 200, (0, 255, 0), 2)
for idb, det in enumerate(detections):
bbox = det.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
# cv2.putText(frame, str(classes[idb]), (int(bbox[0]), int(bbox[1])), 0, 5e-3 * 100, (0, 255, 0), 2)
# cv2.imshow('gallery', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
list_file.write(str(frame_index)+' ')
if len(boxs) != 0:
for i in range(0,len(boxs)):
list_file.write(str(boxs[i][0]) + ' '+str(boxs[i][1]) + ' '+str(boxs[i][2]) + ' '+str(boxs[i][3]) + ';')
list_file.write('\n')
fps = ( fps + (1./(time.time()-t1)) ) / 2
id_frame +=1
print("idx_frame-%d, fps= %f"%(id_frame, fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
def main():
counter = []
writeVideo_flag = False
display = False
fps = 0.0
info_cam = load_list_video(list_video_path, id_path)
max_cosine_distance = 0.8
nn_budget = 100
nms_max_overlap = 1.0
for info in info_cam:
path = os.path.join(video_path, info[1])
ROI, MOI = load_roi_moi(zones_path, mois_path, info[1])
frame_delay = load_delay('../Dataset_A/time_delay.txt', info[1])
print('delay', frame_delay)
print('MOI', MOI)
car_class = OT(1, 'Car')
truck_class = OT(2, 'Truck')
objects = [car_class, truck_class]
print("Processing video: ", info)
video_capture = cv2.VideoCapture(path)
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
out = cv2.VideoWriter('result_' + info[1] + '.avi', fourcc,
info[-1], (w, h))
pause_display = False
frame_num = 17900
data = []
start_video = time.time()
while True:
start = time.time()
video_capture.set(cv2.CAP_PROP_POS_FRAMES, frame_num)
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
# print(count1)
#print(frame.shape)
w = int(video_capture.get(3))
h = int(video_capture.get(4))
result = []
t1 = time.time()
img = letterbox(frame, new_shape=img_size)[0]
img = img[:, :, ::-1].transpose(2, 0,
1) # BGR to RGB, to 3x416x416
img = np.ascontiguousarray(img)
res = frame.copy()
dets = run_detect(model, img, device, frame)
for det, ob in zip(dets, objects):
ob.predict_obtracker(frame, det)
ob.update_obtracker()
#frame = ob.visualize(frame)
res, data = ob.tracking_ob1(ROI, MOI, frame, info[0],
frame_num, data, frame_delay)
print('saved', len(data))
#frame_num += 1
if display:
draw_roi(ROI, res)
cv2.imshow('frame', res)
if not pause_display:
key = cv2.waitKey(10)
if key == ord('q'):
break
if key == ord(' '):
pause_display = not pause_display
frame_num += 1
else:
key = cv2.waitKey(10)
if key == ord('q'):
break
if key == ord(' '):
pause_display = not pause_display
if writeVideo_flag:
#save a frame
out.write(frame)
print('frame_num: {} fps: {} '.format(frame_num,
(1 / (time.time() - start))))
print('Process' + info[0] + ' :{} h',
format((time.time() - start_video) / 60))
write_result_file(data)
print('wrote')
print(" ")
print("[Finish]")
video_capture.release()
if writeVideo_flag:
out.release()
#list_file.close()
result_file.close()
cv2.destroyAllWindows()
def main(yolo):
t0 = time.time()
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
counter = []
#deep_sort
model_filename = 'model_data/market1501.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1) #跟踪使用
find_objects = ['person']
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
video_capture = cv2.VideoCapture(args["input"])
if writeVideo_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
if args["ids"] == False:
out = cv2.VideoWriter('./output/output%s.avi' % args["camera"][1],
fourcc, 50, (w, h))
else:
out = cv2.VideoWriter(
'./output/output%s_reid.avi' % args["camera"][1], fourcc, 50,
(w, h))
list_file = open('detection_rslt.txt', 'w')
frame_index = -1
nump = 1
#fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
frame2 = copy.deepcopy(frame)
#image = Image.fromarray(frame)
image = Image.fromarray(frame[..., ::-1]) #bgr to rgb 仅yolo使用
boxs, confidence, class_names = yolo.detect_image(image)
print(boxs)
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [
Detection(bbox, 1.0, feature)
for bbox, feature in zip(boxs, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
i = int(0)
indexIDs = []
c = []
boxes = []
makequery = True
for det in detections:
bbox = det.to_tlbr()
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
indexIDs.append(int(track.track_id))
counter.append(int(track.track_id))
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255),
2) #跟踪框
color = [int(c) for c in COLORS[indexIDs[i] % len(COLORS)]]
list_file.write(str(frame_index) + ',') #3-5-7-9
list_file.write(str(track.track_id) + ',') #画面内的所有人id
b0 = str(bbox[0]
) #.split('.')[0] + '.' + str(bbox[0]).split('.')[0][:1]
b1 = str(bbox[1]
) #.split('.')[0] + '.' + str(bbox[1]).split('.')[0][:1]
b2 = str(bbox[2] - bbox[0]
) #.split('.')[0] + '.' + str(bbox[3]).split('.')[0][:1]
b3 = str(bbox[3] - bbox[1])
#放置id
list_file.write(
str(b0) + ',' + str(b1) + ',' + str(b2) + ',' + str(b3))
list_file.write('\n')
if len(class_names) > 0:
class_name = class_names[0]
cv2.putText(frame, str(class_names[0]),
(int(bbox[0]), int(bbox[1] - 20)), 0, 5e-3 * 150,
(255, 255, 255), 2) #person
i += 1
center = (int(
((bbox[0]) + (bbox[2])) / 2), int(((bbox[1]) + (bbox[3])) / 2))
pts[track.track_id].append(center)
thickness1 = 5
for j in range(1, len(pts[track.track_id])):
if pts[track.track_id][j - 1] is None or pts[
track.track_id][j] is None:
continue
thickness = int(np.sqrt(64 / float(j + 1)) * 2)
cv2.line(frame, (pts[track.track_id][j - 1]),
(pts[track.track_id][j]), (255, 255, 255), thickness)
if args["ids"] == False:
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (color), 3)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1] - 50)), 0, 5e-3 * 150,
(color), 2) #id
cv2.circle(frame, (center), 1, color, thickness1)
for j in range(1, len(pts[track.track_id])):
if pts[track.track_id][j - 1] is None or pts[
track.track_id][j] is None:
continue
thickness = int(np.sqrt(64 / float(j + 1)) * 2)
cv2.line(frame, (pts[track.track_id][j - 1]),
(pts[track.track_id][j]), (color), thickness)
try:
num = (int(args["camera"][1]) - 1) * 200
path = 'Z:\\pro2\\whole\\person\\gallery\\%04d' % int(
track.track_id + num)
if not os.path.exists(path):
os.makedirs(path)
if len(os.listdir(path)) <= 150: #最多存储150张相片
crop = frame2[int(bbox[1]):int(bbox[3]),
int(bbox[0]):int(bbox[2])]
crop = cv2.resize(crop, (64, 128),
interpolation=cv2.INTER_AREA
) #CUBIC 对扩大图片 area 对缩小图片
filepath = path + '\\' + '%04d' % int(
track.track_id +
num) + '_%s_' % args["camera"] + '%04d' % int(
len(os.listdir(path)) +
1) + '_%.2f' % (video_capture.get(0) /
1000) + '.jpg' #%04d
cv2.imwrite(filepath, crop)
except:
continue
#单独索引
else:
makequery = False
id1 = int(args["ids"])
if int(track.track_id) == id1:
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (color), 3)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1] - 50)), 0,
5e-3 * 150, (color), 2) #id
cv2.circle(frame, (center), 1, color, thickness1)
for j in range(1, len(pts[track.track_id])):
if pts[track.track_id][j - 1] is None or pts[
track.track_id][j] is None:
continue
thickness = int(np.sqrt(64 / float(j + 1)) * 2)
cv2.line(frame, (pts[track.track_id][j - 1]),
(pts[track.track_id][j]), (color), thickness)
cv2.putText(frame, str(class_names[0]),
(int(bbox[0]), int(bbox[1] - 20)), 0,
5e-3 * 150, (color), 2) #person
else:
continue
count = len(set(counter))
cv2.putText(frame, "Total Pedestrian Counter: " + str(count),
(int(20), int(80)), 0, 5e-3 * 200, (0, 255, 0), 2)
cv2.putText(frame, "Current Pedestrian Counter: " + str(i),
(int(20), int(40)), 0, 5e-3 * 200, (0, 255, 0), 2)
cv2.namedWindow("YOLO4_Deep_SORT", 0)
cv2.resizeWindow('YOLO4_Deep_SORT', 1024, 768)
cv2.imshow('YOLO4_Deep_SORT', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
out.write(frame)
frame_index = frame_index + 1
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
#makequery
if makequery == True:
root_path = 'Z:\\pro2\\whole\\person\\gallery\\'
copy_path = 'Z:\\pro2\\whole\\person\\query\\'
ids = os.listdir(root_path)
#print(ids)
for i in ids:
img_path = root_path + i
img = os.listdir(img_path)
indeximg = img[int(len(img) / 2)]
old_name = img_path + '\\' + indeximg
new_path = copy_path + i
new_name = new_path + '\\' + indeximg
if not os.path.exists(new_path):
os.makedirs(new_path)
shutil.copyfile(old_name, new_name)
print(" ")
print("[Finish]")
end = time.time()
print("the whole time ", end - t0)
if len(pts[track.track_id]) != None:
print(args["input"][43:57] + ": " + str(count) + " " +
str(class_name) + ' Found')
else:
print("[No Found]")
video_capture.release()
if writeVideo_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
def main():
global frame, frame_index, out, list_file, track, count
start = time.time()
# 参数定义
max_cosine_distance = 0.5 # 0.9 余弦距离的控制阈值
nn_budget = None
nms_max_overlap = 0.3 # 非极大抑制的阈值
# 是否保存识别结果
write_video_flag = True
counter = []
# load our serialized model from disk
# print("[INFO] loading model...")
net = cv2.dnn.readNetFromCaffe(args["prototxt"], args["model"])
# deep_sort
model_filename = 'model_data/market1501.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
video_capture = cv2.VideoCapture(args["input"])
obj_count_txt_filename = 'counter.txt'
count_file = open(obj_count_txt_filename, 'a')
count_file.write('\n')
if write_video_flag:
# Define the codec and create VideoWriter object
w = int(video_capture.get(3))
h = int(video_capture.get(4))
# DIVX, XVID, MJPG, X264, WMV1, WMV2.(XVID is more preferable.MJPG results in high size ideo.X264 gives ery small size video)
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(
os.path.join('video',
str(args['input'].split('.')[0][-7:]) + '_out.avi'),
fourcc, 20, (w, h))
list_file = open('detection.txt', 'w')
frame_index = -1
# 帧率计数
fps = 0.0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if not ret:
print("Can't receive frame (stream end?). Exiting ...")
break
time1 = time.time()
# frame = imutils.resize(frame, width=800)
(h, w) = frame.shape[:2]
blob = cv2.dnn.blobFromImage(cv2.resize(frame, (300, 300)), 0.007843,
(300, 300), 127.5)
# predictions 检测
time2 = time.time()
net.setInput(blob)
detections = net.forward()
# detections.shape
# >>> (1, 1, n, 7)
# eg:(1, 1, 2, 7)
# [[[[0. 9. 0.42181703 0.4647404 0.610577
# 0.6360997 0.8479532]
# [0. 15. 0.8989926 0.21603307 0.42735672
# 0.58441484 0.8699994]]]]
boxs = []
class_names = []
for i in np.arange(0, detections.shape[2]):
confidence = detections[0, 0, i, 2]
# greater than the minimum confidence
if confidence > args["confidence"]:
idx = int(detections[0, 0, i, 1])
class_name = CLASSES[idx]
# 筛选类别
if class_name in NEED_CLASSES:
class_names.append(class_name)
box = detections[0, 0, i, 3:7] * np.array([w, h, w, h])
# 转为整形坐标
(startX, startY, endX, endY) = box.astype("int")
startX = 0 if startX < 0 else startX
startY = 0 if startY < 0 else startY
boxs.append([startX, startY, endX - startX, endY - startY])
print(boxs, class_names)
time3 = time.time()
print('detect cost is', time3 - time2)
# 特征提取
features = encoder(frame, boxs)
# score to 1.0 here).
detections = [
Detection(bbox, class_name, 1.0, feature)
for bbox, class_name, feature in zip(boxs, class_names, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
time4 = time.time()
print('features extract is', time4 - time3)
# Call the tracker
tracker.predict()
tracker.update(detections)
time5 = time.time()
print('update tracker cost:', time5 - time4)
i = 0
# 跟踪器id
indexIDs = []
for track in tracker.tracks:
# todo and or
if not track.is_confirmed() or track.time_since_update > 1:
continue
# boxes.append([track[0], track[1], track[2], track[3]])
indexIDs.append(track.track_id)
counter.append(track.track_id)
bbox = track.to_tlbr()
start_x, start_y, end_x, end_y = bbox.astype('int')
color = COLORS[indexIDs[i] % len(COLORS)].tolist()
if not track.flag and track.class_name == 'person':
track.flag = handle_face_car('person', start_x, start_y, end_x,
end_y)
else:
track.flag = handle_face_car(track.class_name, start_x,
start_y, end_x, end_y,
not track.flag)
# 画目标跟踪框、id标注
cv2.rectangle(frame, (start_x, start_y), (end_x, end_y), color, 3)
cv2.putText(frame, track.class_name + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 40)), 0, 0.75, color, 2)
i += 1
# 画运动轨迹 draw motion path
center = int(((bbox[0]) + (bbox[2])) / 2), int(
((bbox[1]) + (bbox[3])) / 2)
pts[track.track_id].append(center)
thickness = 5
cv2.circle(frame, center, 1, color, thickness)
for j in range(1, len(pts[track.track_id])):
if pts[track.track_id][j - 1] is None or pts[
track.track_id][j] is None:
continue
thickness = int(np.sqrt(64 / (j + 1.0)) * 2)
cv2.line(frame, (pts[track.track_id][j - 1]),
(pts[track.track_id][j]), color, thickness)
time6 = time.time()
print('handle tracker cost:', time6 - time5)
# 画目标检测白框
# for det in detections:
# bbox = det.to_tlbr()
# cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
count = len(set(counter))
cv2.putText(frame, "Total Object Counter: " + str(count), (20, 120), 0,
0.75, (0, 255, 0), 2)
cv2.putText(frame, "Current Object Counter: " + str(i), (20, 80), 0,
0.75, (0, 255, 0), 2)
cv2.putText(frame, "FPS: %f" % fps, (20, 40), 0, 1.0, (0, 255, 0), 2)
# time7 = time.time()
# print('Draw Rectangle and Text cost:', time7 - time6)
cv2.namedWindow("SSD_Deep_SORT", 0)
cv2.resizeWindow('SSD_Deep_SORT', 1024, 768)
cv2.imshow('SSD_Deep_SORT', frame)
if write_video_flag:
# save a frame
out.write(frame)
frame_index += 1
list_file.write(str(frame_index) + ' ')
if len(boxs) != 0:
for i in range(0, len(boxs)):
list_file.write(
str(boxs[i][0]) + ' ' + str(boxs[i][1]) + ' ' +
str(boxs[i][2]) + ' ' + str(boxs[i][3]) + ' ')
list_file.write('\n')
fps = (fps + (1. / (time.time() - time1))) / 2
# print(set(counter))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
print("[Finish]")
end = time.time()
# if len(pts[track.track_id]):
# print(str(args["input"]) + ": " + str(count) + 'target Found')
# count_file.write(str("[VIDEO]: " + args["input"]) + " " + (
# str(count)) + " " + "[MODEL]: MobileNetSSD" + " " + "[TIME]:" + (str('%.2f' % (end - start))))
# else:
# print("[No Found]")
video_capture.release()
count_file.write('\n')
count_file.close()
if write_video_flag:
out.release()
list_file.close()
cv2.destroyAllWindows()
#! /usr/bin/env python
## Program to find the prime numbers from 1 to n.
import math
from timeit import time
primeno = []
def prime(n):
for i in xrange(2, n + 1):
sq = math.sqrt(i)
count = 0
for j in primeno:
if j > sq + 1:
break
if i % j == 0:
count += 1
break
if count == 0:
primeno.append(i)
start_time = time.time()
prime(10**5) # You can specify any number
end_time = time.time()
total_time = end_time - start_time # Time in seconds
#print primeno # uncomment this line if you want to see the prime numbers.
print 'Total time:', total_time
def main():
salir = False
while salir == False:
print("Seleccione lo que desea ejecutar: \n")
print("1: Bezout.")
print("2: Inverso modular. ")
print("3: Potencia modular. ")
print("4: Test Miller-Rabin. ")
print("5: Paso enano-gigante. ")
print("6: Teorema chino de los restos. ")
print("7: Jacobi. ")
print("8: Raices modulares. ")
print("9: Pollard. ")
print("10: Fermat. ")
print("11: Levantar raices. ")
print("12: salir")
seleccion = input()
if seleccion == "1":
#5,9
print("##################BEZOUT################## \n")
print("Dime el valor de a: \n")
a = int(input())
print("Dime el valor de b: \n")
b = int(input())
start_time = time.time()
salida = Bezout(a,b)
elapsed_time = time.time() - start_time
print("Los valores de Bezout son:","\nValor de u: ",salida[0],"\nValor de v: ",salida[1],"\nValor de mcd: ",salida[2])
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "2":
print("##################INVERSO MODULAR################## \n")
print("Dime el valor de a: \n")
a = int(input())
print("Dime el valor de b: \n")
b = int(input())
start_time = time.time()
salida1 = inverso_modular(a,b)
elapsed_time = time.time() - start_time
print("El inverso modular es: " , salida1)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "3":
print("##################POTENCIA MODULAR################## \n")
print("Dime el valor de a: \n")
a = int(input())
print("Dime el valor de n: \n")
n = int(input())
print("Dime el valor de p: \n")
p = int(input())
start_time = time.time()
salida2 = potencia_mod(a,n,p)
elapsed_time = time.time() - start_time
print("Potencia Modular: ",salida2)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "4":
print("##################TEST DE MILLER-RABIN################## \n")
print("Dime el valor de a: \n")
a = int(input())
start_time = time.time()
salida3 = test_Miller_Rabin(a)
elapsed_time = time.time() - start_time
print("Resultado Test Miller-Rabin: ",salida3)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "5":
print("##################PASO ENANO-GIGANTE################## \n")
print("Dime el valor de a: \n")
a = int(input())
print("Dime el valor de n: \n")
n = int(input())
print("Dime el valor de p: \n")
p = int(input())
start_time = time.time()
salida4 = Paso_enano_gigante(a,b,p)
elapsed_time = time.time() - start_time
print("La solución Paso-enano-gigante es: ",salida4)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "6":
print("##################TEOREMA CHINO DE LOS RESTOS################## \n")
print("Dime el valor de a: \n")
a = int(input())
print("Dime el valor de n: \n")
n = int(input())
print("Dime el valor de p: \n")
p = int(input())
print("Dime el valor de q: \n")
q = int(input())
start_time = time.time()
salida5 = T_chino_restos(a,b,p,q)
elapsed_time = time.time() - start_time
print("La salida del Teorema chino de los restos es: ",salida5)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "7":
print("##################JACOBI################## \n")
print("Dime el valor de p: \n")
p = int(input())
print("Dime el valor de m: \n")
m = int(input())
start_time = time.time()
salida6 = Jacobi(p,m)
elapsed_time = time.time() - start_time
print("La salida de Jacobi es: ", salida6)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "8":
print("##################RAICES MODULARES################## \n")
print("Dime el valor de a: \n")
a = int(input())
print("Dime el valor de p: \n")
p = int(input())
start_time = time.time()
salida7 = raices_modulares(a,p)
elapsed_time = time.time() - start_time
print("La raiz modular es: ",salida7)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "9":
print("##################POLLARD################## \n")
print("Dime el valor de a: \n")
a = int(input())
start_time = time.time()
salida8 = Pollard(a)
elapsed_time = time.time() - start_time
print("Pollard es: ",salida8)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "10":
print("##################FERMAT################## \n")
print("Dime el valor de a: \n")
a = int(input())
start_time = time.time()
salida9 = Fermat(91)
elapsed_time = time.time() - start_time
print("Fermat es:",salida9)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "11":
print("##################LEVANTAR RAICES################## \n")
print("Dime el valor de a: \n")
a = int(input())
print("Dime el valor de p: \n")
p = int(input())
print("Dime el valor de q: \n")
q = int(input())
start_time = time.time()
salida10 = levantar_raices(a,p,q)
elapsed_time = time.time() - start_time
print("levantar raices: ",salida10)
print("Elapsed time: %0.10f seconds." % elapsed_time)
elif seleccion == "12":
salir = True
else:
print("Selección incorrecta \n")
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.2
nn_budget = None
nms_max_overlap = 1.0
output_format = 'mp4'
video_name = 'bus4_2in_4out.mp4'
file_path = join('data_files/videos', video_name)
output_name = 'save_data/out_' + video_name[0:-3] + output_format
initialize_door_by_yourself = False
door_array = None
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
show_detections = True
writeVideo_flag = True
asyncVideo_flag = False
counter = Counter(counter_in=0, counter_out=0, track_id=0)
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(output_name, fourcc, 15, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
ret, first_frame = video_capture.read()
if door_array is None:
if initialize_door_by_yourself:
door_array = select_object(first_frame)[0]
print(door_array)
else:
all_doors = read_door_info('data_files/doors_info.csv')
door_array = all_doors[video_name]
border_door = door_array[3]
error_values = []
truth = get_truth(video_name)
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if not ret:
total_count = counter.return_total_count()
true_total = truth.inside + truth.outside
err = abs(total_count - true_total) / true_total
log_res = "in video: {}\n predicted / true\n counter in: {} / {}\n counter out: {} / {}\n" \
" total: {} / {}\n error: {}\n______________\n".format(video_name, counter.counter_in,
truth.inside,
counter.counter_out, truth.outside,
total_count, true_total, err)
with open('log_results.txt', 'w') as file:
file.write(log_res)
print(log_res)
error_values.append(err)
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls,
feature) for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
cv2.rectangle(frame, (int(door_array[0]), int(door_array[1])),
(int(door_array[2]), int(door_array[3])), (23, 158, 21),
2)
for det in detections:
bbox = det.to_tlbr()
if show_detections and len(classes) > 0:
score = "%.2f" % (det.confidence * 100) + "%"
rect_head = Rectangle(bbox[0], bbox[1], bbox[2], bbox[3])
rect_door = Rectangle(int(door_array[0]), int(door_array[1]),
int(door_array[2]), int(door_array[3]))
intersection = rect_head & rect_door
if intersection:
squares_coeff = rect_square(*intersection) / rect_square(
*rect_head)
cv2.putText(
frame,
score + " inter: " + str(round(squares_coeff, 3)),
(int(bbox[0]), int(bbox[3])), 0, 1e-3 * frame.shape[0],
(0, 100, 255), 5)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 3)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
# first appearence of object with id=track.id
if track.track_id not in counter.people_init or counter.people_init[
track.track_id] == 0:
counter.obj_initialized(track.track_id)
rect_head = Rectangle(bbox[0], bbox[1], bbox[2], bbox[3])
rect_door = Rectangle(door_array[0], door_array[1],
door_array[2], door_array[3])
res = rect_head & rect_door
if res:
inter_square = rect_square(*res)
head_square = rect_square(*rect_head)
# was initialized in door, probably going in
if (inter_square / head_square) >= 0.8:
counter.people_init[track.track_id] = 2
# initialized in the bus, mb going out
elif (inter_square /
head_square) <= 0.4 or bbox[3] > border_door:
counter.people_init[track.track_id] = 1
# res is None, means that object is not in door contour
else:
counter.people_init[track.track_id] = 1
counter.people_bbox[track.track_id] = bbox
counter.cur_bbox[track.track_id] = bbox
adc = "%.2f" % (track.adc *
100) + "%" # Average detection confidence
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 1e-3 * frame.shape[0],
(0, 255, 0), 5)
if not show_detections:
track_cls = track.cls
cv2.putText(frame, str(track_cls),
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.putText(
frame, 'ADC: ' + adc,
(int(bbox[0]), int(bbox[3] + 2e-2 * frame.shape[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
id_get_lost = [
track.track_id for track in tracker.tracks
if track.time_since_update >= 25 and track.age >= 29
]
id_inside_tracked = [
track.track_id for track in tracker.tracks if track.age > 60
]
for val in counter.people_init.keys():
# check bbox also
cur_c = find_centroid(counter.cur_bbox[val])
init_c = find_centroid(counter.people_bbox[val])
vector_person = (cur_c[0] - init_c[0], cur_c[1] - init_c[1])
if val in id_get_lost and counter.people_init[val] != -1:
# if vector_person < 0 then current coord is less than initialized, it means that man is going
# in the exit direction
if vector_person[1] > 70 and counter.people_init[
val] == 2: # and counter.people_bbox[val][3] > border_door \
counter.get_in()
elif vector_person[1] < -70 and counter.people_init[val] == 1:
counter.get_out()
counter.people_init[val] = -1
print(f"person left frame")
print(f"current centroid - init : {cur_c} - {init_c}\n")
print(f"vector: {vector_person}\n")
del val
# elif val in id_inside_tracked and val not in id_get_lost and counter.people_init[val] == 1 \
# and bb_intersection_over_union(counter.cur_bbox[val], door_array) <= 0.3 \
# and vector_person[1] > 0: # and \
# # counter.people_bbox[val][3] > border_door:
# counter.get_in()
#
# counter.people_init[val] = -1
# print(f"person is tracked for a long time")
# print(f"current centroid - init : {cur_c} - {init_c}\n")
# print(f"vector: {vector_person}\n")
# imaggg = cv2.line(frame, find_centroid(counter.cur_bbox[val]),
# find_centroid(counter.people_bbox[val]),
# (0, 0, 255), 7)
# cv2.imshow('frame', imaggg)
# cv2.waitKey(0)
ins, outs = counter.show_counter()
cv2.putText(frame, "in: {}, out: {} ".format(ins, outs), (10, 30), 0,
1e-3 * frame.shape[0], (255, 0, 0), 5)
cv2.namedWindow('image', cv2.WINDOW_NORMAL)
cv2.resizeWindow('image', 1400, 800)
cv2.imshow('image', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
# print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
mean_error = np.mean(error_values)
print("mean error for {} video: {}".format(video_name, mean_error))
res = DE.DE()
# print(model.depen(res))
return res
def tuner(model, data):
if model == rforest:
return _de(tuneRF, data)
elif model == CART:
return _de(tuneCART, data)
if __name__ == '__main__':
from timeit import time
data = explore(dir='../Data/')[0][5] # Only training data to tune.
print(data)
# set_trace()
for m in [tuneRF]:
t = time.time()
mdl = m(data)
# _test(data)
tunings = _de(m, data)
print(tunings)
print(mdl.depen(tunings))
print(time.time() - t)
# print _de()
# print main()
# import sk; xtile = sk.xtile
# print xtile(G)
# main(dir = 'Data/')
def main():
start = time.time()
counter = []
writeVideo_flag = False
fps = 0.0
filename_path = os.path.join(result_path, 'submission.txt')
list_video, list_ids = load_list_video(list_video_path, id_path)
result_file = open(filename_path, 'w')
max_cosine_distance=0.8
nn_budget = 100
nms_max_overlap = 1.0
display = True
for video in list_video:
path = os.path.join(video_path, video)
ROI = load_roi(zones_path, video)
vis=visualization.Visualization(img_shape=(960,1280,3), update_ms=2000)
metric = nn_matching.NearestNeighborDistanceMetric(
"cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
results = []
print("Processing video: ", video )
video_capture = cv2.VideoCapture(path)
pause_display = False
frame_num = 0
while True:
start = time.time()
# print(count)
video_capture.set(cv2.CAP_PROP_POS_FRAMES, frame_num)
ret, frame = video_capture.read() # frame shape 640*480*3
# gray = cv2.cvtColor(frame1, cv2.COLOR_BGR2GRAY)
# frame = np.zeros_like(frame1)
# frame[:,:,0] = gray
# frame[:,:,1] = gray
# frame[:,:,2] = gray
if ret != True:
break
# print(count1)
#print(frame.shape)
w = int(video_capture.get(3))
h = int(video_capture.get(4))
result = []
t1 = time.time()
img = letterbox(frame, new_shape=img_size)[0]
img = img[:, :, ::-1].transpose(2, 0, 1) # BGR to RGB, to 3x416x416
img = np.ascontiguousarray(img)
dets = run_detect(model,img,device,frame)
detectionss=[]
for det in dets:
feature = gdet.HOG_feature(frame, det[:4])
detectionss.append(Detection(det[:4], det[4], feature, det[-1]))
#detectionss.append(Detection(det[:4], det[4], det[-1]) for det in dets)
img = np.zeros((h, w, 3), np.uint8)
img = frame.copy()
min_confidence = 0.4
detections = [d for d in detectionss if d.confidence >= min_confidence]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(
boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Update tracker.
tracker.predict()
tracker.update(detections)
if display:
vis.set_image(frame.copy())
vis.draw_detections(detections)
vis.draw_trackers(tracker.tracks)
res = vis.return_img()
draw_roi(ROI, res)
cv2.imshow('frame', res)
print('frame_num', frame_num)
if not pause_display:
key = cv2.waitKey(2)
if key == ord('q'):
break
if key == ord(' '):
pause_display = not pause_display
frame_num += 1
else:
key = cv2.waitKey(0)
if key == ord('q'):
break
if key == ord(' '):
pause_display = not pause_display
print(" ")
print("[Finish]")
video_capture.release()
if writeVideo_flag:
out.release()
#list_file.close()
result_file.close()
cv2.destroyAllWindows()
def update(self,frame,recent_pupil_positions,events):
for pt in recent_pupil_positions:
#if the eye is detected
if pt['norm_gaze'] is not None:
self.nb_frame += 1
#if the eye was closed
if(self.eye_open == False):
self.time_eye_o = time.time()
self.eye_open = True
#BLINK CONTROL
#if the eye is opened too long, previous blinks are forgotten
if(time.time() - self.time_eye_o) > self.FORGET_BLINK:
self.nb_blink = 0
#if the previous closed session was long, nb_blink is incremented
if(time.time() - self.time_eye_c) > self.BLINK:
self.nb_blink += 1
print "Nb blink :"
print self.nb_blink
#if there are 2 blinks : we switch the control mode (0->1 and 1->0)
if(self.nb_blink >= 2):
if(self.control_mode.value == self.OFF):
self.control_mode.value = self.ON
else:
self.control_mode.value = self.OFF
self.nb_blink = 0
print "Mode control"
print self.control_mode.value
self.time_eye_c = time.time()
#STATE CONTROL
if (self.nb_frame > self.STATE_WAIT and self.control_mode.value == self.ON):
#self.LEFT
if (pt['norm_gaze'][0] < self.LEFT_BORDER) and (self.state.value!=self.BACKWARD):
self.previous_state = self.state.value
self.state.value = self.LEFT
self.nb_frame = 0
elif (self.state.value == self.LEFT) and (pt['norm_gaze'][0] > self.LEFT_BORDER):
if self.previous_state == self.STOP:
self.state.value = self.STOP
self.nb_frame = 0
else:
self.state.value = self.NORMAL_SPEED
self.nb_frame = 0
#self.RIGHT
elif (pt['norm_gaze'][0] > self.RIGHT_BORDER) and (self.state.value!=self.BACKWARD):
self.previous_state = self.state.value
self.state.value = self.RIGHT
self.nb_frame = 0
elif (self.state.value == self.RIGHT) and (pt['norm_gaze'][0] < self.RIGHT_BORDER):
if self.previous_state == self.STOP:
self.state.value = self.STOP
self.nb_frame = 0
else:
self.state.value = self.NORMAL_SPEED
self.nb_frame = 0
#NORMAL SPEED
elif (pt['norm_gaze'][1]>self.UP_BORDER) and (self.state.value==self.STOP):
self.state.value = self.NORMAL_SPEED
self.nb_frame = 0
elif (pt['norm_gaze'][1]<self.DOWN_BORDER) and (self.state.value==self.HIGH_SPEED):
self.state.value = self.NORMAL_SPEED
self.nb_frame = 0
#HIGH SPEED
elif (pt['norm_gaze'][1]>self.UP_BORDER) and (self.state.value==self.NORMAL_SPEED):
self.state.value = self.HIGH_SPEED
self.nb_frame = 0
#self.BACKWARD
elif (pt['norm_gaze'][1]<self.DOWN_BORDER) and (self.state.value==self.STOP):
self.state.value = self.BACKWARD
self.nb_frame = 0
#self.STOP
elif (pt['norm_gaze'][1]>self.UP_BORDER) and (self.state.value==self.BACKWARD):
self.state.value = self.STOP
self.nb_frame = 0
elif (pt['norm_gaze'][1]<self.DOWN_BORDER) and (self.state.value==self.NORMAL_SPEED):
self.state.value = self.STOP
self.nb_frame = 0
#END if nb_frame > STATE_WAIT
self.gaze_x.value = pt['norm_gaze'][0]
self.gaze_y.value = pt['norm_gaze'][1]
self.pupil_display_list.append(pt['norm_gaze'])
#if the eye is not detected
else:
#if the eye was open
if self.eye_open == True:
self.time_eye_c = time.time()
self.eye_open = False
self.time_eye_o = time.time()
#if the eye is closed too long, we stop
if (time.time() - self.time_eye_c) > self.BLINK:
self.state.value = self.STOP
print "Volunteer blink"
self.pupil_display_list[:-3] = []
if self.window_should_close:
self.close_window()
if self.window_should_open:
self.open_window()
