def main():
cap = cv2.VideoCapture('goose.mp4')
tracker = Tracker(160, 30, 5, 1)
pause = False
firstFrame = None
counted = []
while cap.isOpened():
ret, frame = cap.read()
if ret:
frame = imutils.resize(frame, width=800)
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
gray = cv2.GaussianBlur(gray, (3, 3), 0)
if firstFrame is None:
firstFrame = gray
continue
frameDelta = cv2.absdiff(firstFrame, gray)
thresh = cv2.threshold(frameDelta, 25, 255, cv2.THRESH_BINARY)[1]
thresh = cv2.dilate(thresh, None, iterations=2)
cnts = cv2.findContours(thresh.copy(), cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
cnts = imutils.grab_contours(cnts)
centers = []
for i, c in enumerate(cnts):
if cv2.contourArea(c) > 500:
continue
(x, y, w, h) = cv2.boundingRect(c)
centers.append(np.array([[x + w / 2], [y + h / 2]]))
cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 1)
if (len(centers) > 0):
tracker.Update(centers)
for track in tracker.tracks:
x = track.prediction[0]
y = track.prediction[1]
id = track.track_id
if x > 400 and id not in counted:
counted.append(id)
cv2.putText(frame, str(id), (x, y),
cv2.FONT_HERSHEY_SIMPLEX, 0.4, (0, 255, 0), 1)
cv2.putText(frame, str(len(counted)), (20, 20),
cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 1)
cv2.imshow('Tracking', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
else:
break
cap.release()
cv2.destroyAllWindows()
def main():
# Create opencv video capture object
cap = cv2.VideoCapture('project2.avi')
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(40, 8, 5, 100)
# Variables initialization
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
out = cv2.VideoWriter('Tracking2_wait8.avi',
cv2.VideoWriter_fourcc(*'DIVX'), 5, (200, 200))
while (True):
ret, frame = cap.read()
if ret == True:
centers = detector.Detect(frame)
# If centroids are detected then track them
if (len(centers) >= 0):
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr], 2)
# Display the resulting tracking frame
cv2.imshow('Tracking', frame)
cv2.waitKey(100)
out.write(frame)
else:
break
cap.release()
cv2.destroyAllWindows()
out.release()
def main():
# Create opencv video capture object
cap = cv2.VideoCapture('G:/cmu/colonoscopy/New folder/Cold.mp4')
#cap = cv2.VideoCapture('G:/cmu/colonoscopy/imagemark/Color-Tracker-master/Retroflect-at-end.mp4')
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(160, 1000, 5, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
num = 0
frame_num = 0
# Infinite loop to process video frames
while (True):
frame_num += 1
print(frame_num)
# Capture frame-by-frame
ret, frame = cap.read()
frame = frame[30:550, 400:930]
#frame = frame[40:400,130:450]
# Make copy of original frame
orig_frame = copy.copy(frame)
# Skip initial frames that display logo
if (skip_frame_count < 15):
skip_frame_count += 1
continue
# Detect and return centeroids of the objects in the frame
centers = detector.Detect1(orig_frame)
# If centroids are detected then track them
if (len(centers) > 0):
text = 'Biopsy'
cv2.putText(orig_frame,
text, (100, 100),
cv2.FONT_HERSHEY_SIMPLEX,
1.0, (0, 0, 255),
2,
lineType=cv2.LINE_AA)
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 2)
# Display the resulting tracking frame
cv2.imshow('Tracking', frame)
# Display the original frame
cv2.imshow('Original', orig_frame)
print(num)
# Slower the FPS
cv2.waitKey(20)
# Check for key strokes
k = cv2.waitKey(50) & 0xff
if k == 27: # 'esc' key has been pressed, exit program.
break
if k == 112: # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if (pause is True):
print("Code is paused. Press 'p' to resume..")
while (pause is True):
# stay in this loop until
key = cv2.waitKey(30) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
corners[3][1] = min(corners[3][1], corners1[3][1])
print(corners1)
continue
corners = np.float32(corners)
# Perspective transform on frame to get table only with right measurements:
frame = getTableFromFrame(corners, frame)
# Detect box centers and angles:
(centers, angles) = detector.Detect(frame)
# Track box centers:
if (len(centers) > 0):
tracker.Update(centers)
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv.line(frame, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 2)
# Display the resulting tracking frame
cv.imshow('Tracking', frame)
def detect_video(yolo, video_path, output_path=""):
vid = cv2.VideoCapture(video_path)
if not vid.isOpened():
raise IOError("Couldn't open webcam or video")
video_FourCC = int(vid.get(cv2.CAP_PROP_FOURCC))
video_fps = vid.get(cv2.CAP_PROP_FPS)
video_size = (int(vid.get(cv2.CAP_PROP_FRAME_WIDTH)),
int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT)))
isOutput = True if output_path != "" else False
if isOutput:
#print("!!! TYPE:", type(output_path), type(video_FourCC), type(video_fps), type(video_size))
out = cv2.VideoWriter(output_path, video_FourCC, video_fps, video_size)
accum_time = 0
curr_fps = 0
fps = "FPS: ??"
prev_time = timer()
tracker = Tracker(160, 30, 6, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
while True:
return_value, frame = vid.read()
print(frame.shape)
image = Image.fromarray(frame)
image,centers,number = yolo.detect_image(image)
print(image.size)
result = np.asarray(image)
curr_time = timer()
exec_time = curr_time - prev_time
prev_time = curr_time
accum_time = accum_time + exec_time
curr_fps = curr_fps + 1
if accum_time > 1:
accum_time = accum_time - 1
fps = "FPS: " + str(curr_fps)
curr_fps = 0
font = cv2.FONT_HERSHEY_SIMPLEX
#cv2.putText(result, text=fps, org=(3, 15), fontFace=cv2.FONT_HERSHEY_SIMPLEX,fontScale=0.50, color=(255, 0, 0), thickness=2)
cv2.putText(result, str(number), (20, 40), font, 1, (0, 0, 255), 5)
#for k in range(len(centers)):
#cv2.circle(frame, centers[k], 3, (255, 255, 0), 3)
#cv2.imshow("xiao",frame)
#cv2.waitKey(100000)
############################################################################################
#print(len(centers))
#for i in range(len(centers)):
#print(centers[i])
#cv2.waitKey(0)
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(result, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 4)
#x3 = tracker.tracks[i].track_id
#cv2.putText(result,str(tracker.tracks[j].track_id),(int(x1),int(y1)),font,track_colors[j],3)
#cv2.circle(result,(int(x1),int(y1)),3,track_colors[j],3)
# Display the resulting tracking frame
cv2.imshow('Tracking', result)
###################################################
cv2.namedWindow("result", cv2.WINDOW_NORMAL)
cv2.imshow("result", result)
if isOutput:
out.write(result)
if cv2.waitKey(100) & 0xFF == ord('q'):
break
yolo.close_session()
def main():
command_time = False
# Get a list of ALL the sticks that are plugged in
# we need at least one
devices = mvnc.EnumerateDevices()
if len(devices) == 0:
print('No devices found')
quit()
# Pick the first stick to run the network
device = mvnc.Device(devices[0])
# Open the NCS
device.OpenDevice()
# The graph file that was created with the ncsdk compiler
# graph_file_name = 'graph_heli'
graph_file_name = 'graph_all'
# read in the graph file to memory buffer
with open(graph_file_name, mode='rb') as f:
graph_in_memory = f.read()
# create the NCAPI graph instance from the memory buffer containing the graph file.
graph = device.AllocateGraph(graph_in_memory)
drone = libardrone.ARDrone()
# drone = False
# cap = cv2.VideoCapture('unobstructed.m4v')
cap = cv2.VideoCapture('tcp://192.168.1.1:5555')
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
center = [width / 2, height / 2]
print ("width " , width)
print ("height " , height)
fps = 0.0
i = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
tracker = Tracker(dist_thresh=50, max_frames_to_skip=5, max_trace_length=10, trackIdCount=100)
detect = True
AUTO = False
while cap.isOpened():
start = time.time()
for j in range(10):
ret = cap.grab()
ret, img = cap.retrieve()
ts = datetime.datetime.utcnow().isoformat()
if ret == True and detect:
i += 1
# fileName = datetime.datetime.utcnow().isoformat().replace(":","").replace("-","").replace(".","");
# cv2.imwrite('images/' + fileName + '.jpg', img)
if (SPLIT):
split_width = 340
split_height = 300
#we do not need to copy the image, as we can pass it by reference to the NCS
#we should make copies if we debug / want to separately redraw detections
img00 = img[0:split_height, 0:split_width]#.copy()
img01 = img[0:split_height, (width-split_width):width]#.copy()
img10 = img[(height-split_height):height, 0:split_width]#.copy()
img11 = img[(height-split_height):height, (width-split_width):width]#.copy()
cv2.imshow('img00', img00)
cv2.imshow('img01', img01)
cv2.imshow('img10', img10)
cv2.imshow('img11', img11)
detectedBoxes = run_inference(img00, graph, split_width, split_height, 0, 0, box_color=box_color00) + \
run_inference(img01, graph, split_width, split_height, width - split_width, 0, box_color=box_color01) + \
run_inference(img10, graph, split_width, split_height, 0, height-split_height, box_color=box_color10) + \
run_inference(img11, graph, split_width, split_height, width - split_width, height-split_height, box_color=box_color11)
else:
detectedBoxes = run_inference(img, graph, width, height, 0, 0, global_box_color)
if (CLUSTERING):
distance = 35
detectedBoxes = cluster(detectedBoxes, distance)
tracker.Update(detectedBoxes)
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace)-1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0]
y1 = tracker.tracks[i].trace[j][1]
x2 = tracker.tracks[i].trace[j+1][0]
y2 = tracker.tracks[i].trace[j+1][1]
clr = tracker.tracks[i].track_id % 9
cv2.line(img, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 2)
for box in detectedBoxes:
cv2.circle(img, (int(box[1]), int(box[2])), distance, (255, 255, 255), thickness=3, lineType=8, shift=0)
if (len(tracker.tracks) > 0):
tracker_width = False
tracker_height = False
# if (tracker.tracks[0].last_detection_assigment is not None and tracker.tracks[0].skipped_frames == 0 and tracker.tracks[0].age > 1):
if (tracker.tracks[0].last_detection_assigment is not None):
tracker_width = detectedBoxes[tracker.tracks[0].last_detection_assigment][3]
tracker_height = detectedBoxes[tracker.tracks[0].last_detection_assigment][4]
if (AUTO):
updatePID(center, tracker.tracks[0].prediction, tracker_width, tracker_height, drone)
elif (AUTO):
drone.turn_left()
end = time.time()
seconds = end - start
fps = 1 / seconds
fpsImg = cv2.putText(img, "%.2f fps" % (fps), (20, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.4, (255, 0, 0), 2)
fpsImg = cv2.putText(img, "AUTO MODE: ", (20, 40), cv2.FONT_HERSHEY_SIMPLEX, 0.4, (255, 0, 0), 2)
auto_mode_color = (0, 255, 0)
if (AUTO):
auto_mode_color = (0, 0, 255)
fpsImg = cv2.putText(img, "%s" % (str(AUTO)), (110, 40), cv2.FONT_HERSHEY_SIMPLEX, 0.4, auto_mode_color, 2)
battery_text_color = (0, 255, 0)
battery_text_font = 0.4
if (drone):
if (drone.navdata[0]['battery'] < 20):
battery_text_color = (0, 0, 255)
battery_text_font = 0.7
fpsImg = cv2.putText(img, "%s battery" % (str(drone.navdata[0]['battery'])), (20, 310), cv2.FONT_HERSHEY_SIMPLEX, battery_text_font, battery_text_color, 2)
cv2.imshow("detected", fpsImg)
key = cv2.waitKey(33)
if key == ord('t'):
command_time = time.time()
drone.takeoff()
if key == ord('l'):
command_time = time.time()
drone.land()
if key == ord('h'):
command_time = time.time()
drone.hover()
if key == ord('r'):
tracker = Tracker(dist_thresh=50, max_frames_to_skip=5, max_trace_length=10, trackIdCount=100)
print ('RESETTING TRACKER')
# left joystick
if key == ord('a'):
command_time = time.time()
drone.move_left()
if key == ord('d'):
command_time = time.time()
drone.move_right()
if key == ord('w'):
command_time = time.time()
drone.move_forward()
if key == ord('s'):
command_time = time.time()
drone.move_backward()
# right joystick
if key == ord(';'):
command_time = time.time()
drone.turn_left()
if key == ord('\\'):
command_time = time.time()
drone.turn_right()
if key == ord('['):
command_time = time.time()
drone.move_up()
if key == ord('\''):
command_time = time.time()
drone.move_down()
if key == ord('z'):
AUTO = not AUTO
print ("AUTO MODE ", AUTO)
if key == ord('q'):
break
if (command_time):
command_age = time.time() - command_time
print (command_age)
if (command_age > 0.7):
drone.hover()
command_time = False
print ('hovering again')
else:
if (drone):
drone.hover()
# Clean up the graph and the device
graph.DeallocateGraph()
device.CloseDevice()
class DataWriter:
def __init__(self,
save_video=False,
savepath='examples/res/1.avi',
fourcc=cv2.VideoWriter_fourcc(*'XVID'),
fps=25,
frameSize=(640, 480),
queueSize=1024,
bg_color=None):
"""
提取骨架信息，合成最后视频
:param save_video: 是否保存合成骨架信息的视频
:param savepath: 保存视频的路径
:param fourcc: 保存视频的格式
:param fps: 保存视频的帧率
:param frameSize: 保存视频的尺寸
:param queueSize:
:param bg_color:保存视频的背景，默认None使用原视频作为背景，white用纯白作为背景，black用黑色作为背景
"""
if save_video:
# initialize the file video stream along with the boolean
# used to indicate if the thread should be stopped or not
self.stream = cv2.VideoWriter(savepath, fourcc, fps, frameSize)
assert self.stream.isOpened(), 'Cannot open video for writing'
self.save_video = save_video
self.stopped = False
self.final_result = []
# initialize the queue used to store frames read from
# the video file
self.Q = Queue(maxsize=queueSize)
if opt.save_img:
if not os.path.exists(opt.outputpath + '/vis'):
os.mkdir(opt.outputpath + '/vis')
self.bg_color = bg_color
self.current_frame = 0
self.current_sec = 0
self.out_img_dir = '0_3'
self.tracker = Tracker(160, 30, 50, 100)
def start(self):
# start a thread to read frames from the file video stream
t = Thread(target=self.update, args=())
t.daemon = True
t.start()
return self
def update(self):
# keep looping infinitely
while True:
# if the thread indicator variable is set, stop the
# thread
if self.stopped:
if self.save_video:
self.stream.release()
return
# otherwise, ensure the queue is not empty
if not self.Q.empty():
(boxes, scores, hm_data, pt1, pt2, orig_img, im_name, shotname,
fps) = self.Q.get()
orig_img = np.array(orig_img, dtype=np.uint8)
if boxes is None:
print('boxes~~~~~~~~~~~~~~~~~~~~~is None')
img = orig_img
if opt.vis:
cv2.imshow("AlphaPose Demo", img)
cv2.waitKey(30)
else:
centers = []
img = orig_img
pedestrians = {}
counter = 0
for i, box in enumerate(boxes):
cv2.putText(img, "i:%s" % i, (box[0], box[1] + 20),
font, 0.6, (0, 255, 0), 1, cv2.LINE_AA)
cv2.rectangle(img, (box[0], box[1]), (box[2], box[3]),
(0, 255, 0), 1)
# center_point = self.center(box)
min_x = min(box[0].int(), box[2].int())
max_x = max(box[0].int(), box[2].int())
min_y = min(box[1].int(), box[3].int())
max_y = max(box[1].int(), box[3].int())
x = (min_x + max_x) / 2
y = (min_y + max_y) / 2
w = max_x - min_x
h = max_y - min_y
cv2.rectangle(img, (box[0], box[1]), (box[2], box[3]),
(0, 255, 0), 1)
# cv2.circle(img, (x, y), 4, (0, 0, 255), -1)
b = np.array([[x], [y]])
centers.append(np.round(b))
counter += 1
# Track object using Kalman Filter
# 使用卡尔曼滤波器跟踪对象
self.tracker.Update(centers)
# For identified object tracks draw tracking line
# 对于已识别的目标轨迹，绘制跟踪线
# Use various colors to indicate different track_id
# 使用不同的颜色表示不同的轨迹ID
for i in range(len(self.tracker.tracks)):
if (len(self.tracker.tracks[i].trace) > 1):
for j in range(
len(self.tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = self.tracker.tracks[i].trace[j][0][0]
y1 = self.tracker.tracks[i].trace[j][1][0]
x2 = self.tracker.tracks[i].trace[j + 1][0][0]
y2 = self.tracker.tracks[i].trace[j + 1][1][0]
clr = self.tracker.tracks[i].track_id % 9
cv2.line(img, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr],
2)
# Display the resulting tracking frame
cv2.imshow('Tracking', img)
# Slower the FPS
cv2.waitKey(50)
else:
time.sleep(0.1)
def running(self):
# indicate that the thread is still running
time.sleep(0.2)
return not self.Q.empty()
def save(self, boxes, scores, hm_data, pt1, pt2, orig_img, im_name,
shotname, fps):
# save next frame in the queue
self.Q.put((boxes, scores, hm_data, pt1, pt2, orig_img, im_name,
shotname, fps))
def stop(self):
# indicate that the thread should be stopped
self.stopped = True
time.sleep(0.2)
def results(self):
# return final result
return self.final_result
def len(self):
# return queue len
return self.Q.qsize()
def main():
fig = plt.figure()
args = config_parse()
imgs = []
graphs = []
if not (args.traj or args.plot):
print("Please choose plot method")
return
if args.traj:
ax = plt.axes(projection='3d')
else:
fig, axs = plt.subplots()
dataset = ['dic', 'fluo', 'phc']
if args.type in dataset:
if args.type == 'dic':
#this associate range when trace backward, used to detect mitosis
# the new trace needs to link old trace in previous frame
minD = 30
maxD = 70
#the dissassociateRange of kalman track
dissassociateRange = 30
if args.type == 'fluo':
raise NotImplementedError
if args.type == 'phc':
raise NotImplementedError
else:
print(f"Please choose from {dataset}")
return
img = plt.imread("mask_DIC/mask000.tif")
size = img.shape
width = size[0]
height = size[1]
#we cant delete
tracker = Tracker(dissassociateRange, 100, 0)
detector = Detector()
l = []
for root, dirs, files in os.walk("mask_DIC"):
for file in files:
l.append(file)
l = sorted(l)
#original file
ori = []
for root, dirs, files in os.walk("Sequence 1"):
for file in files:
ori.append(file)
ori = sorted(ori)
first = -1
second = 0
#center list
Clist = []
plt_trace = []
for name in l:
imgN = os.path.join("mask_DIC/" + name)
print(imgN)
img = plt.imread(imgN)
graph = Graph(img, height, width)
centers = detector.Detect(graph)
Clist.append(centers)
tracker.Update(centers, second)
if args.traj:
#TODO repair traj hasn't been done
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
# for j in range(len(tracker.tracks[i].trace)-1):
x1 = tracker.tracks[i].trace[-2][0]
y1 = tracker.tracks[i].trace[-2][1]
x2 = tracker.tracks[i].trace[-1][0]
y2 = tracker.tracks[i].trace[-1][1]
if x1 == x2 and y1 == y2:
continue
ax.plot([x1, x2], [y1, y2], [first, second])
plt.draw()
if args.plot:
oriN = os.path.join("Sequence 1/" + ori[second])
originImg = plt.imread(oriN)
imgs.append(originImg)
plt_trace.append(copy.deepcopy(tracker.tracks))
graphs.append(copy.deepcopy(graph))
first += 1
second += 1
if second == args.num:
break
plt_trace = MitosisRecovery(tracker, plt_trace, minD, maxD)
if args.plot:
ani = FuncAnimation(fig,
update,
fargs=(imgs, axs, Clist, plt_trace, graphs),
interval=args.interval,
frames=second)
plt.show()
if args.save:
ani.save('myAnimation.gif', writer='imagemagick', fps=15)
done = args.search
while done:
choice = input("> ")
qlist = ['speed', 'total', 'net']
if choice in qlist:
id = int(input("ID: "))
frame = int(input("which frame u are: "))
if frame < 1:
print("the frame has to be great than 0")
continue
t = tracker.findTrack(id)
if t == None:
print("Dont have this cell")
continue
if len(t.frame) == 1:
print("Sorry, this cell only appear once")
absolute_val_array = np.abs(np.array(t.frame) - frame)
smallest_difference_inx = absolute_val_array.argmin()
closet_frame = t.frame[smallest_difference_inx]
if closet_frame != frame:
print(
f"Sorry we can't find {id} in frame {frame}, the closet frame is {closet_frame}"
)
if choice == "speed":
pre_frame = t.frame[closet_frame - 1]
pre_loc = np.array(t.trace[smallest_difference_inx - 1])
cur_loc = np.array(t.trace[smallest_difference_inx])
dist = np.linalg.norm(cur_loc - pre_loc)
speed = dist / (frame - pre_frame)
print(
f"The {id} at frame {frame} has a speed {speed} pixel/frame"
)
if choice == "total":
t.printTrace(smallest_difference_inx)
dist = t.totalDistance(smallest_difference_inx)
print(f"It has travelled {dist} in total")
if choice == "net":
loc = np.array(t.trace[smallest_difference_inx])
start = np.array(t.trace[0])
dist = np.linalg.norm(loc - start)
print(f"The net distance is {dist}")
if choice == "q":
done = False
for iii in range (0,len(df)):
if df.iat[iii,3] == (frameaa+1):
if df.iat[iii,2] in [3,6,8]:
bb = np.array([[df.iat[iii,0]],[df.iat[iii,1]]])
centers.append(np.round(bb))
iii = iii + 1
# If centroids are detected then track them
if (len(centers) > 0):
print ('frame number=',framea)
# Track object using Kalman Filter
centers.sort(key=lambda x:x[1])
tracker.Update(centers,frameaa+1)
cv.putText(frame, str(frameaa+1),(20,20),cv.FONT_HERSHEY_DUPLEX,0.8,(255,255,255))
for icc in range (0,len(centers)):
cv.circle(frame, (centers[icc][0], centers[icc][1]), 5, (0, 255, 0), 2)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
print ('trackid=',tracker.tracks[i].track_id)
trackidcheck = tracker.tracks[i].track_id
if (len(tracker.tracks[i].trace) > 10):
#------------------------------------------------------------------------------------------------
tracks_dataframe = pd.DataFrame(columns = ['trackID','x','y','tcolor','framenum','detectedx','detectedy'])
def detect_video(yolo, video_path, output_path=""):
vid = cv2.VideoCapture(video_path)
if not vid.isOpened():
raise IOError("Couldn't open webcam or video")
video_FourCC = int(vid.get(cv2.CAP_PROP_FOURCC))
video_fps = vid.get(cv2.CAP_PROP_FPS)
video_size = (int(vid.get(cv2.CAP_PROP_FRAME_WIDTH)),
int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT)))
isOutput = True if output_path != "" else False
if isOutput:
out = cv2.VideoWriter(output_path, video_FourCC, video_fps, video_size)
accum_time = 0
curr_fps = 0
prev_time = timer()
# Variables initialization
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
# 初始化一个tracker， 用来管理Tracks
tracker = Tracker(160, 30, 6, 100)
while True:
# 读取视频文件，得到每一帧图像
return_value, frame = vid.read()
image = Image.fromarray(frame)
# 利用yolo检测
image, centers, number = yolo.detect_image(image)
result = np.asarray(image)
curr_time = timer()
exec_time = curr_time - prev_time
prev_time = curr_time
accum_time = accum_time + exec_time
curr_fps = curr_fps + 1
if accum_time > 1:
accum_time = accum_time - 1
fps = "FPS: " + str(curr_fps)
curr_fps = 0
cv2.putText(result, str(number), (20, 40), cv2.FONT_HERSHEY_SIMPLEX, 1,
(0, 0, 255), 5)
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers)
# 画出跟踪轨迹
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(result, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr], 4)
cv2.imshow('Tracking', result)
cv2.namedWindow("result", cv2.WINDOW_NORMAL)
cv2.imshow("result", result)
if isOutput:
out.write(result)
if cv2.waitKey(100) & 0xFF == ord('q'):
break
yolo.close_session()
def main():
# Load model and run graph inception resnet v1 from models and file resnetv1_inception.py
tf.reset_default_graph()
sess = tf.InteractiveSession()
sess = tf.Session()
images_pl = tf.placeholder(tf.float32, shape=[None, 160, 160, 3], name='input_image')
images_norm = tf.map_fn(lambda frame: tf.image.per_image_standardization(frame), images_pl)
train_mode = tf.placeholder(tf.bool)
age_logits, gender_logits, _ = inception_resnet_v1.inference(images_norm, keep_probability=0.8,
phase_train=train_mode,
weight_decay=1e-5)
gender = tf.argmax(tf.nn.softmax(gender_logits), 1)
age_ = tf.cast(tf.constant([i for i in range(0, 101)]), tf.float32)
age = tf.reduce_sum(tf.multiply(tf.nn.softmax(age_logits), age_), axis=1)
init_op = tf.group(tf.global_variables_initializer(),
tf.local_variables_initializer())
sess.run(init_op)
saver = tf.train.Saver()
ckpt = tf.train.get_checkpoint_state("/home/neosai/Documents/projects/deep_face_recognition/weights/models_gender_and_age/")
if ckpt and ckpt.model_checkpoint_path:
saver.restore(sess, ckpt.model_checkpoint_path)
print("restore model!")
else:
pass
print("error")
# faces = np.empty((1,160, 160, 3))
aligned_images = []
# upscale_factor = 4
# model_name = "/home/neosai/Documents/projects/deep_face_recognition/weights/netG_epoch_4_100.pth"
# # limit_mem()
# model = Generator(upscale_factor).eval()
# # model.cuda()
# model.load_state_dict(torch.load(model_name, map_location=lambda storage, loc: storage))
# Choose area of interest
get_crop_size(path)
# get_crop_size1(path)
print('Your area of interest: ', ix, ' ', iy, ' ', ex, ' ', ey)
area = (ix, iy, ex, ey)
# mid_y = (iy + ey) / 2
# print('point line: ', ix1, ' ', iy1, ' ', ix2, ' ', iy2)
# point1 = (ix, ey)
# point2 = (ex, iy)
# y = ax + b
# a = float((iy2 - iy1) / (ix2 - ix1))
# b = iy2 - a * ix2
# print("a,, b: ", a, b)
# Create opencv video capture object
cap = cv2.VideoCapture(path)
w = int(cap.get(3))
h = int(cap.get(4))
if cap_from_stream:
w = 1280
h = 720
# fourcc = cv2.VideoWriter_fourcc(*'MJPG')
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('../video/1_0222.avi', fourcc, 15, (w, h))
# Create Object Detector
detector = YOLO()
# Create Object Tracker
# tracker = Tracker(iou_thresh=0.3, max_frames_to_skip=5, max_trace_length=20, trackIdCount=0)
tracker = Tracker(iou_thresh=0.1, max_frames_to_skip=10, max_trace_length=20, trackIdCount=0)
# Variables initialization
# track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
# (0, 255, 255), (255, 0, 255), (255, 127, 255),
# (127, 0, 255), (127, 0, 127)]
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
# count_vehicle = {'person': 0, 'motorbike': 0, 'car': 0, 'truck': 0, 'bicycle': 0, 'bus': 0}
# count_people = {'female_0_10':0, 'female_10_20': 0, 'female_20_30': 0,
# 'female_30_40': 0, , 'female_40_50': 0, 'female_50_100': 0
# 'male_0_10': 0, 'male_10_20': 0, 'male_20_30': 0, 'male_30_40': 0, 'male_40_50': 0, 'male_50_100': 0}
count_people_come_in_out = {'female_come_in': 0, 'female_come_out': 0, 'male_come_in': 0, 'male_come_out': 0}
count_people = {'people_come_in': 0, 'people_come_out': 0}
id_number = 0
while cap.isOpened():
# Capture frame-by-frame
ret, frame = cap.read()
if cap_from_stream:
frame = cv2.resize(frame, (1280, 720))
frame = Image.fromarray(frame)
# Detect and return centeroids of the objects in the frame
result, centers, box_detected, obj_type = detector.detect_image(frame, area)
result = np.asarray(result)
frame = np.asarray(frame)
#####
# for bbox in box_detected:
# a0, a1, a2, a3 = bbox[0], bbox[1], bbox[2], bbox[3]
# cv2.rectangle(result, (a0 - 10, a1 - 10), (a2 + 10, a3 + 10), (0, 255, 0), 3)
# if a1 < 480 + 50 and a1 > 480 -50:
# print(a1)
# image_crop = frame[a1:a3, a0:a2]
# # cv2.imwrite("image.jpg", image_crop)
# image_crop = Image.fromarray(image_crop, 'RGB')
# # image_crop = super_resolution_image(image_crop, model)
# image_crop_array = np.asarray(image_crop)
# face_male_resize = image_crop.resize((160, 160), Image.ANTIALIAS)
# face = np.array(face_male_resize)
# aligned_images.append(face)
# # faces[0, :, :, :] = face
# age_predict, gender_predict = sess.run([age, gender], feed_dict={images_pl: aligned_images, train_mode: False})
# aligned_images = []
# # print(gender_predict)
# # print(type(gender_predict))
# label = "{}, {}".format(int(age_predict[0]), "Female" if gender_predict[0] == 0 else "Male")
# id_number += 1
# print(label)
# name = "../image/102/id_{}, {}".format(id_number, label)
# cv2.imwrite(name + ".jpg", image_crop_array)
# cv2.rectangle(result, (a0 - 5, a1 - 5), (a2 + 5, a3 + 5), color=(0, 0, 255),
# thickness=3)
# cv2.putText(result, label, (a0 + 6, a1 - 6), font, 2, (0, 255, 0), 3, cv2.LINE_AA)
#####
# print('Number of detections: ', len(centers))
# a = 0
# If centroids are detected then track them
print("len box_detected: ", len(box_detected))
if len(box_detected) > 0:
# Track object using Kalman Filter
tracker.Update(box_detected, obj_type)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
# print("trace of track i: ",len(tracker.tracks[i].trace))
# print("len tracker: ", len(tracker.tracks[i].trace))
# if len(tracker.tracks[i].trace) == 0:
# bbox = tracker.tracks[i].ground_truth_box.reshape((4, 1))
# a0, a1, a2, a3 = convert_bbox(bbox)
# image_crop = frame[a1:a3, a0:a2]
# cv2.imwrite("image.jpg", image_crop)
# image_crop = Image.fromarray(image_crop, 'RGB')
# image_crop = super_resolution_image(image_crop, model)
# image_crop_array = np.asarray(image_crop)
# face_male_resize = image_crop.resize((160, 160), Image.ANTIALIAS)
# face = np.array(face_male_resize)
# aligned_images.append(face)
# # faces[0, :, :, :] = face
# age_predict, gender_predict = sess.run([age, gender], feed_dict={images_pl: aligned_images, train_mode: False})
# aligned_images = []
# # print(gender_predict)
# # print(type(gender_predict))
# label = "{}, {}".format(int(age_predict[0]), "Female" if gender_predict[0] == 0 else "Male")
# id_number += 1
# print(label)
# name = "../image/103/id_{}, {}".format(id_number, label)
# cv2.imwrite(name + ".jpg", image_crop_array)
# cv2.rectangle(result, (a0 - 5, a1 - 5), (a2 + 5, a3 + 5), color=(0, 0, 255),
# thickness=3)
# cv2.putText(result, label, (a0 + 6, a1 - 6), font, 2, (0, 255, 0), 3, cv2.LINE_AA)
# if len(tracker.tracks[i].trace) >= 9:
if len(tracker.tracks[i].trace) > 0:
if len(tracker.tracks[i].trace) == 1:
print("a")
bbox = tracker.tracks[i].ground_truth_box.reshape((4, 1))
a0, a1, a2, a3 = convert_bbox(bbox)
image_crop = frame[a1:a3, a0:a2]
cv2.imwrite("image.jpg", image_crop)
image_crop = Image.fromarray(image_crop, 'RGB')
# image_crop = super_resolution_image(image_crop, model)
image_crop_array = np.asarray(image_crop)
face_male_resize = image_crop.resize((160, 160), Image.ANTIALIAS)
face = np.array(face_male_resize)
aligned_images.append(face)
# faces[0, :, :, :] = face
age_predict, gender_predict = sess.run([age, gender], feed_dict={images_pl: aligned_images, train_mode: False})
aligned_images = []
# print(gender_predict)
# print(type(gender_predict))
label = "{}, {}".format(int(age_predict[0]), "Female" if gender_predict[0] == 0 else "Male")
tracker.tracks[i].age = str(int(age_predict[0]))
tracker.tracks[i].gender = "Female" if gender_predict[0] == 0 else "Male"
id_number += 1
print(label)
name = "../image/102/id_{}, {}".format(id_number, label)
cv2.imwrite(name + ".jpg", image_crop_array)
cv2.rectangle(result, (a1, a0), (a3, a2), color=(255, 0, 0),
thickness=1)
cv2.putText(result, label, (a1 + 6, a0 - 6), font, 2, (255, 0, 0), 1, cv2.LINE_AA)
# x_center_first = tracker.tracks[i].trace[0][0][0]
# y_center_first = tracker.tracks[i].trace[0][1][0]
if (len(tracker.tracks[i].trace) > 1):
bbox = tracker.tracks[i].ground_truth_box.reshape((4, 1))
a0, a1, a2, a3 = convert_bbox(bbox)
label = "{}, {}".format(tracker.tracks[i].age, tracker.tracks[i].gender)
cv2.rectangle(result, (a0, a1), (a2, a3), color=(0, 0, 255),
thickness=1)
cv2.putText(result, label, (a0 + 6, a1 - 6), font, 2, (0, 0, 255), 1, cv2.LINE_AA)
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(result, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 2)
# cv2.putText(result, label, ())
# classes = tracker.tracks[i].get_obj()
if (len(tracker.tracks[i].trace) >= 9) and (not tracker.tracks[i].counted):
bbox = tracker.tracks[i].ground_truth_box.reshape((4, 1))
tracker.tracks[i].counted = True
x_center_first = tracker.tracks[i].trace[0][0][0]
y_center_first = tracker.tracks[i].trace[0][1][0]
x_center_second = tracker.tracks[i].trace[7][0][0]
y_center_second = tracker.tracks[i].trace[7][1][0]
# if y_center_first > (a * x_center_first + b):
# count_people["people_come_out"] += 1
# if y_center_first < (a * x_center_first + b):
# count_people["people_come_in"] += 1
if y_center_second > y_center_first and x_center_second > x_center_first:
count_people["people_come_in"] += 1
if y_center_second < y_center_first and x_center_second < x_center_first:
count_people["people_come_out"] += 1
# a0, a1, a2, a3 = convert_bbox(bbox)
# image_crop = frame[a1:a3, a0:a2]
# # cv2.imwrite("image.jpg", image_crop)
# image_crop = Image.fromarray(image_crop, 'RGB')
# image_crop = super_resolution_image(image_crop, model)
# image_crop_array = np.asarray(image_crop)
# face_male_resize = image_crop.resize((160, 160), Image.ANTIALIAS)
# face = np.array(face_male_resize)
# aligned_images.append(face)
# # faces[0, :, :, :] = face
# age_predict, gender_predict = sess.run([age, gender], feed_dict={images_pl: aligned_images, train_mode: False})
# aligned_images = []
# # print(gender_predict)
# # print(type(gender_predict))
# label = "{}, {}".format(int(age_predict[0]), "Female" if gender_predict[0] == 0 else "Male")
# id_number += 1
# print(label)
# name = "../image/102/id_{}, {}".format(id_number, label)
# cv2.imwrite(name + ".jpg", image_crop_array)
# cv2.rectangle(result, (a0, a1), (a2, a3), color=(255, 0, 0),
# thickness=3)
# cv2.putText(result, label, (a0 + 6, a1 - 6), font, 2, (0, 255, 0), 3, cv2.LINE_AA)
# Display the resulting tracking frame
x = 30
y = 30
dy = 20
i = 0
font = cv2.FONT_HERSHEY_COMPLEX_SMALL
for key, value in count_people.items():
text = key + ':' + str(value)
cv2.putText(result, text, (x, y + dy * i), font, 1, (255, 0, 255), 2, cv2.LINE_AA)
i += 1
# cv2.line(result, (ix1, iy1), (ix2, iy2), (0, 0, 255), 2)
cv2.circle(result, (ix1, iy1), 5, (0, 0, 255), 4)
cv2.rectangle(result, (ix, iy), (ex, ey), (0, 255, 0), 0)
cv2.imshow('Tracking', result)
out.write(result)
# Check for key strokes
k = cv2.waitKey(1) & 0xff
if k == ord('n'):
continue
elif k == 27: # 'esc' key has been pressed, exit program.
break
# When everything done, release the capture
out.release()
cap.release()
cv2.destroyAllWindows()
def main():
# Create opencv video capture object
cap = cv2.VideoCapture('/home/deepak/innovation_lab_files/vid1_new.mp4')
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(160, 30, 5, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
# Infinite loop to process video frames
mainlist = [[None, None]] * 1000
CarCount = 0
NoneCarCount = 0
NoneVehicle = 0
while (True):
# Capture frame-by-frame
ret, frame = cap.read()
# Make copy of original frame
orig_frame = copy.copy(frame)
# Skip initial frames that display logo
if (skip_frame_count < 15):
skip_frame_count += 1
continue
# Detect and return centeroids of the objects in the frame
centers = detector.Detect(frame)
newcenter = []
# If centroids are detected then track them
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
# print(len(tracker.tracks))
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 4):
# print(tracker.tracks[i].trace)
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 2)
cv2.putText(frame, str(tracker.tracks[i].track_id),
(int(x1), int(y1)), cv2.FONT_HERSHEY_SIMPLEX,
0.75, (0, 0, 255), 2)
newcenter.append(
[int(x1), int(y1), tracker.tracks[i].track_id])
# Display the resulting tracking frame
# cv2.line(frame,(0,0),(100,100),(22,122,222),8)
cv2.line(frame, (200, 600), (960, 600), (139, 0, 0), 8)
cv2.putText(frame, 'Car Count =' + str(CarCount), (30, 30),
cv2.FONT_HERSHEY_SIMPLEX, 0.75, (92, 142, 215), 3)
cv2.putText(frame, 'Non Car Count =' + str(NoneCarCount),
(300, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.75,
(92, 142, 215), 3)
print(CarCount + NoneCarCount)
# cv2.line(frame,(150,450),(280,370),(139,0,0),8)
cv2.imshow('Tracking', frame)
for j in range(len(centers)):
for i in range(len(newcenter)):
a = newcenter[i][0]
b = newcenter[i][1]
e = newcenter[i][2]
c = centers[j][0][0]
d = centers[j][1][0]
temp_len = np.sqrt((a - c) * (a - c) + (b - d) * (b - d))
if (temp_len < 7):
if (mainlist[e][0] != None):
c = mainlist[e][0]
d = mainlist[e][1]
if ((d <= 600) and (c >= 200) and (c <= 960)
and (a >= 200) and (a <= 960) and b >= 600):
CarCount += 1
s1 = orig_frame.shape[0]
s2 = orig_frame.shape[1]
# print('this')
# print(s1)
# print(s2)
# print(a)
# print(b)
# print('this')
# if((a-120>=0) and (a+120<=s2) and (b-120>=0) and (b+120<=s1)):
try:
img = orig_frame[a - 80:a + 80, b - 80:b + 80]
# cv2.imshow("cropped", img)
img = cv2.resize(img, (img_width, img_height))
arr = np.array(img).reshape(
(3, img_width, img_height))
arr = np.expand_dims(arr, axis=0)
prediction = model.predict(arr)[0]
# print(prediction)
bestclass = ''
bestconf = -1
best = [
'non-vehicle', 'vehicle', 'non-vehicle',
'non-vehicle', 'non-vehicle',
'non-vehicle', 'non-vehicle',
'non-vehicle', 'non-vehicle', 'vehicle'
]
for n in [0, 1, 2]:
if (prediction[n] > bestconf):
bestclass = n
bestconf = prediction[n]
if (bestclass != 1 and bestclass != 9):
NoneVehicle += 1
if (NoneVehicle % 10 == 2):
CarCount -= 1
NoneCarCount += 1
# else :
except:
print('this is already vehicle')
mainlist[e][0] = a
mainlist[e][1] = b
else:
mainlist[e][0] = a
mainlist[e][1] = b
newcenter.pop(i)
break
# for i in range(len(newcenter)):
# mainlist[newcenter[i][2]][0]=newcenter[i][0]
# mainlist[newcenter[i][2]][1]=newcenter[i][1]
# Display the original frame
# cv2.imshow('Original', orig_frame)
# Slower the FPS
cv2.waitKey(50)
# Check for key strokes
k = cv2.waitKey(50) & 0xff
if k == 27: # 'esc' key has been pressed, exit program.
break
if k == 112: # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if (pause is True):
print("Code is paused. Press 'p' to resume..")
while (pause is True):
# stay in this loop until
key = cv2.waitKey(30) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
# When everything done, release the capture
print("this is final car count ")
print(CarCount)
cap.release()
cv2.destroyAllWindows()
def detect_video(yolo, video_path, output_path=""):
vid = cv2.VideoCapture(1)
if not vid.isOpened():
raise IOError("Couldn't open webcam or video")
video_FourCC = int(vid.get(cv2.CAP_PROP_FOURCC))
video_fps = vid.get(cv2.CAP_PROP_FPS)
video_size = (int(vid.get(cv2.CAP_PROP_FRAME_WIDTH)),
int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT)))
isOutput = True if output_path != "" else False
if isOutput:
#print("!!! TYPE:", type(output_path), type(video_FourCC), type(video_fps), type(video_size))
out = cv2.VideoWriter(output_path, video_FourCC, video_fps, video_size)
accum_time = 0
curr_fps = 0
fps = "FPS: ??"
prev_time = timer()
tracker = Tracker(160, 1, 9, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
select_area = SelectArea()
area_list = []
while True:
return_value, matrix = vid.read()
# # 选择区域
# if len(area_list) == 0:
# select_area.run(matrix)
# if len(select_area.area_list) <= 0:
# if select_area.select_times < 5:
# continue
# else:
# exit(0)
# # end if
# # end if
# area_list = select_area.area_list
# temp_height = select_area.area_list[0]["height"]
# temp_width = select_area.area_list[0]["width"]
# # # 字体重新初始化
# # draw.set_font(temp_height//9, temp_height//27, (temp_width//20, temp_height//20), temp_height)
# # # 输出视频put流重新初始化
# # camera.width = temp_width
# # camera.height = temp_height
# select_area.clear()
# x1, y1, x2, y2 = area_list[0]["box"]
# frame = matrix[y1:y2, x1:x2]
frame = matrix
print(frame.shape)
image = Image.fromarray(frame)
image,centers,number = yolo.detect_image(image)
print(image.size)
result = np.asarray(image)
curr_time = timer()
exec_time = curr_time - prev_time
prev_time = curr_time
accum_time = accum_time + exec_time
curr_fps = curr_fps + 1
if accum_time > 1:
accum_time = accum_time - 1
fps = "FPS: " + str(curr_fps)
curr_fps = 0
font = cv2.FONT_HERSHEY_SIMPLEX
#cv2.putText(result, text=fps, org=(3, 15), fontFace=cv2.FONT_HERSHEY_SIMPLEX,fontScale=0.50, color=(255, 0, 0), thickness=2)
# cv2.putText(result, str(number), (20, 40), font, 1, (0, 0, 255), 5)
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(result, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 4)
#x3 = tracker.tracks[i].track_id
#cv2.putText(result,str(tracker.tracks[j].track_id),(int(x1),int(y1)),font,track_colors[j],3)
#cv2.circle(result,(int(x1),int(y1)),3,track_colors[j],3)
# Display the resulting tracking frame
cv2.imshow('Tracking', result)
###################################################
cv2.namedWindow("result", cv2.WINDOW_NORMAL)
cv2.imshow("result", result)
if isOutput:
out.write(result)
if cv2.waitKey(100) & 0xFF == ord('q'):
break
yolo.close_session()
def main():
"""Main function for multi object tracking
Usage:
$ python2.7 objectTracking.py
Pre-requisite:
- Python2.7
- Numpy
- SciPy
- Opencv 3.0 for Python
Args:
None
Return:
None
"""
# Create opencv video capture object
cap = cv2.VideoCapture(
'C:\Users\user\Documents\Iot-Tracking\CV2\kalman_filter_multi_object_tracking-master\data\RAW_ Moment van mows down pedestrians in Barcelona caught on camera (DISTURBING FOOTAGE).mp4'
)
#cap = cv2.VideoCapture(0)
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(160, 30, 5, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
# Infinite loop to process video frames
while (True):
# Capture frame-by-frame
ret, frame = cap.read()
# Make copy of original frame
orig_frame = copy.copy(frame)
# Skip initial frames that display logo
if (skip_frame_count < 15):
skip_frame_count += 1
continue
# Detect and return centeroids of the objects in the frame
centers = detector.Detect(frame)
# If centroids are detected then track them
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
#cv2.line(frame, (int(x1), int(y1)), (int(x2), int(y2)), track_colors[clr], 2)
# Display the resulting tracking frame
cv2.imshow('Tracking', frame)
# Display the original frame
cv2.imshow('Original', orig_frame)
# Slower the FPS
cv2.waitKey(50)
# Check for key strokes
k = cv2.waitKey(1) & 0xff
if k == ord("q"): # 'esc' key has been pressed, exit program.
break
if k == ord(
"p"): # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if (pause is True):
print("Code is paused. Press 'p' to resume..")
while (pause is True):
# stay in this loop until
key = cv2.waitKey(1) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
def main():
# Create opencv video capture object
# cap = cv2.VideoCapture('data/TrackingBugs.mp4')
cap = cv2.VideoCapture('data/video_3_bin.mp4')
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(160, 30, 5, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
frame_count = 1
# Infinite loop to process video frames
while True:
# Capture frame-by-frame
ret, frame = cap.read()
# Make copy of original frame
orig_frame = copy.copy(frame)
# Convert binary image to greyscale
frame[frame != 0] = 255
# Skip initial frames that display logo
if skip_frame_count < 1:
skip_frame_count += 1
continue
# Detect and return centroids of the objects in the frame
if ret:
print "Processing frame " + format(frame_count)
frame_count += 1
centers = detector.Detect(frame)
# If centroids are detected then track them
if len(centers) > 0:
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if len(tracker.tracks[i].trace) > 1:
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr], 2)
# Display the resulting tracking frame
cv2.imshow('Tracking', frame)
# Display the original frame
cv2.imshow('Original', orig_frame)
# Slower the FPS
cv2.waitKey(50)
# # # Check for key strokes
# k = cv2.waitKey(50) & 0xff
# if k == 27: # 'esc' key has been pressed, exit program.
# break
# if k == 112: # 'p' has been pressed. this will pause/resume the code.
# pause = not pause
# if pause is True:
# print "Code is paused. Press 'p' to resume.."
# while pause is True:
# # stay in this loop until
# key = cv2.waitKey(30) & 0xff
# if key == 112:
# pause = False
# print "Resume code..!!"
# break
else:
print "All frames were processed"
break
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
def main():
tf.reset_default_graph()
sess = tf.InteractiveSession()
sess = tf.Session()
aligned_images = []
upscale_factor = 4
img_size = 64
depth = 16
k = 8
weight_file = "/home/neosai/Documents/github/age-gender-estimation/utkface/weights.29-3.76_utk.hdf5"
model_predict_age_and_gender = WideResNet(img_size, depth=depth, k=k)()
model_predict_age_and_gender.load_weights(weight_file)
model_name = "/home/neosai/Documents/projects/deep_face_recognition/weights/netG_epoch_4_100.pth"
model = Generator(upscale_factor).eval()
model.load_state_dict(
torch.load(model_name, map_location=lambda storage, loc: storage))
get_crop_size(path)
get_crop_size1(path)
print('Your area of interest: ', ix, ' ', iy, ' ', ex, ' ', ey)
area = (ix, iy, ex, ey)
print("iy1", iy1)
# Create opencv video capture object
cap = cv2.VideoCapture(path)
w = int(cap.get(3))
h = int(cap.get(4))
if cap_from_stream:
w = 1280
h = 720
# fourcc = cv2.VideoWriter_fourcc(*'MJPG')
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(
'/home/neosai/Documents/projects/deep_face_recognition/video/1_02222.avi',
fourcc, 15, (w, h))
# Create Object Detector
detector = YOLO()
tracker = Tracker(iou_thresh=0.3,
max_frames_to_skip=5,
max_trace_length=40,
trackIdCount=0)
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
count_people_come_in_out = {
'female_come_in': 0,
'female_come_out': 0,
'male_come_in': 0,
'male_come_out': 0
}
count_people = {'people_come_in': 0, 'people_come_out': 0}
id_number = 0
img_size = 64
font = cv2.FONT_HERSHEY_COMPLEX_SMALL
# depth = 16
# k = 8
# weight_file = "/home/neosai/Documents/github/age-gender-estimation/utkface/weights.29-3.76_utk.hdf5"
# model_predict_age_and_gender = WideResNet(img_size, depth=depth, k=k)()
# model_predict_age_and_gender.load_weights(weight_file)
faces = np.empty((1, 64, 64, 3))
while cap.isOpened():
# Capture frame-by-frame
ret, frame = cap.read()
if cap_from_stream:
frame = cv2.resize(frame, (1280, 720))
frame = Image.fromarray(frame)
# Detect and return centeroids of the objects in the frame
result, centers, box_detected, obj_type = detector.detect_image(
frame, area)
result = np.asarray(result)
frame = np.asarray(frame)
#####
for bbox in box_detected:
a0, a1, a2, a3 = bbox[0], bbox[1], bbox[2], bbox[3]
cv2.rectangle(result, (a0 - 10, a1 - 10), (a2 + 10, a3 + 10),
(0, 255, 0), 3)
print(a1)
print(iy1)
if a1 < iy1 + 50 and a1 > iy1 - 50:
image_crop = frame[a1:a3, a0:a2]
cv2.imwrite("image.jpg", image_crop)
image_crop = Image.fromarray(image_crop, 'RGB')
image_crop = super_resolution_image(image_crop, model)
image_crop_array = np.asarray(image_crop)
face_male_resize = image_crop.resize((img_size, img_size),
Image.ANTIALIAS)
face = np.array(face_male_resize)
aligned_images.append(face)
faces[0, :, :, :] = face
# age_predict, gender_predict = sess.run([age, gender], feed_dict={images_pl: aligned_images, train_mode: False})
# predict ages and genders of the detected faces
results = model_predict_age_and_gender.predict(faces)
predicted_genders = results[0]
ages = np.arange(0, 101).reshape(101, 1)
predicted_ages = results[1].dot(ages).flatten()
aligned_images = []
label = "{}, {}".format(
int(predicted_ages[0]),
"F" if predicted_genders[0][0] > 0.5 else "M")
# print(gender_predict)
# print(type(gender_predict))
# label = "{}, {}".format(int(age_predict[0]), "Female" if gender_predict[0] == 0 else "Male")
id_number += 1
print(label)
name = "../image/102/id_{}, {}".format(id_number, label)
cv2.imwrite(name + ".jpg", image_crop_array)
cv2.rectangle(result, (a0 - 5, a1 - 5), (a2 + 5, a3 + 5),
color=(0, 0, 255),
thickness=3)
cv2.putText(result, label, (a0 + 6, a1 - 6), font, 2,
(0, 255, 0), 3, cv2.LINE_AA)
#####
# print('Number of detections: ', len(centers))
# a = 0
# If centroids are detected then track them
if len(box_detected) > 0:
# Track object using Kalman Filter
tracker.Update(box_detected, obj_type)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
# print("trace of track i: ",len(tracker.tracks[i].trace))
# print("len tracker: ", len(tracker.tracks[i].trace))
# if len(tracker.tracks[i].trace) == 0:
# bbox = tracker.tracks[i].ground_truth_box.reshape((4, 1))
# a0, a1, a2, a3 = convert_bbox(bbox)
# image_crop = frame[a1:a3, a0:a2]
# cv2.imwrite("image.jpg", image_crop)
# image_crop = Image.fromarray(image_crop, 'RGB')
# # image_crop = super_resolution_image(image_crop, model)
# image_crop_array = np.asarray(image_crop)
# face_male_resize = image_crop.resize((160, 160), Image.ANTIALIAS)
# face = np.array(face_male_resize)
# aligned_images.append(face)
# # faces[0, :, :, :] = face
# age_predict, gender_predict = sess.run([age, gender], feed_dict={images_pl: aligned_images, train_mode: False})
# aligned_images = []
# # print(gender_predict)
# # print(type(gender_predict))
# label = "{}, {}".format(int(age_predict[0]), "Female" if gender_predict[0] == 0 else "Male")
# id_number += 1
# print(label)
# name = "../image/id_{}, {}".format(id_number, label)
# cv2.imwrite(name + ".jpg", image_crop_array)
# cv2.rectangle(result, (a0 - 5, a1 - 5), (a2 + 5, a3 + 5), color=(0, 0, 255),
# thickness=3)
# cv2.putText(result, label, (a0 + 6, a1 - 6), font, 2, (0, 255, 0), 3, cv2.LINE_AA)
if len(tracker.tracks[i].trace) >= 0:
print(len(tracker.tracks[i].trace))
# cv2.circle(result, (int(x_center_first), int(y_center_first)), 5, (0, 0, 255), -1)
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(result, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr], 2)
classes = tracker.tracks[i].get_obj()
if (len(tracker.tracks[i].trace) >=
9) and (not tracker.tracks[i].counted):
bbox = tracker.tracks[i].ground_truth_box.reshape(
(4, 1))
tracker.tracks[i].counted = True
x_center_first = tracker.tracks[i].trace[0][0][0]
y_center_first = tracker.tracks[i].trace[0][1][0]
x_center_second = tracker.tracks[i].trace[8][0][0]
y_center_second = tracker.tracks[i].trace[8][1][0]
# if y_center_first > (a * x_center_first + b):
# count_people["people_come_out"] += 1
# if y_center_first < (a * x_center_first + b):
# count_people["people_come_in"] += 1
if y_center_second > y_center_first and x_center_second > x_center_first:
count_people["people_come_in"] += 1
if y_center_second < y_center_first and x_center_second < x_center_first:
count_people["people_come_out"] += 1
# a0, a1, a2, a3 = convert_bbox(bbox)
# image_crop = frame[a1:a3, a0:a2]
# cv2.imwrite("image.jpg", image_crop)
# image_crop = Image.fromarray(image_crop, 'RGB')
# image_crop = super_resolution_image(image_crop, model)
# image_crop_array = np.asarray(image_crop)
# face_male_resize = image_crop.resize((160, 160), Image.ANTIALIAS)
# face = np.array(face_male_resize)
# aligned_images.append(face)
# # faces[0, :, :, :] = face
# age_predict, gender_predict = sess.run([age, gender], feed_dict={images_pl: aligned_images, train_mode: False})
# aligned_images = []
# # print(gender_predict)
# # print(type(gender_predict))
# label = "{}, {}".format(int(age_predict[0]), "Female" if gender_predict[0] == 0 else "Male")
# id_number += 1
# print(label)
# name = "../image/id_{}, {}".format(id_number, label)
# cv2.imwrite(name + ".jpg", image_crop_array)
# cv2.rectangle(result, (a0, a1), (a2, a3), color=(255, 0, 0),
# thickness=3)
# cv2.putText(result, label, (a0 + 6, a1 - 6), font, 2, (0, 255, 0), 3, cv2.LINE_AA)
# Display the resulting tracking frame
x = 30
y = 30
dy = 20
i = 0
font = cv2.FONT_HERSHEY_COMPLEX_SMALL
for key, value in count_people.items():
text = key + ':' + str(value)
cv2.putText(result, text, (x, y + dy * i), font, 1, (255, 0, 255),
2, cv2.LINE_AA)
i += 1
# cv2.line(result, (ix1, iy1), (ix2, iy2), (0, 0, 255), 2)
cv2.circle(result, (ix1, iy1), 5, (0, 0, 255), 4)
cv2.rectangle(result, (ix, iy), (ex, ey), (0, 255, 0), 0)
cv2.imshow('Tracking', result)
out.write(result)
# Check for key strokes
k = cv2.waitKey(1) & 0xff
if k == ord('n'):
continue
elif k == 27: # 'esc' key has been pressed, exit program.
break
# When everything done, release the capture
out.release()
cap.release()
cv2.destroyAllWindows()
centroids = []
for contour in contours:
moment = cv2.moments(contour)
if moment[
"m00"] > args.contour: # objects with area more than area threshold detected
if args.debug:
cv2.drawContours(
frame, contours, index, GREEN, 2
) # draw contour in frame, draw only if detected under contour threshold
x = int(moment["m10"] / moment["m00"])
y = int(moment["m01"] / moment["m00"])
center = (x, y) # calculate center of contour
centroids.append((contour, center, moment["m00"], 0))
index += 1
tracker.Update(centroids, (count / fps))
fishes = tracker.fishes
# exit program if there are no fishes detected
if len(fishes) is 0:
print("No fish detected")
sys.exit()
total_avg_speed = 0
total_area = 0
for i in range(len(fishes)):
fish = tracker.fishes[i]
center = fish.centroid
if args.debug:
def detect_img_series(yolo, img_path, output_path=""):
imgs = os.listdir(img_path)
imgs.sort(key=lambda x: int(x[:-4])) # 按照名字进行排序
accum_time = 0
curr_fps = 0
prev_time = timer()
# Variables initialization
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
# 初始化一个tracker， 用来管理Tracks
tracker = Tracker(160, 30, 6, 100)
for img in imgs:
image = cv2.imread(img_path + '/' + img)
image = Image.fromarray(image)
# 利用yolo检测
image, centers, number = yolo.detect_image(image)
result = np.asarray(image)
curr_time = timer() #获取当前时刻时间
exec_time = curr_time - prev_time
prev_time = curr_time
accum_time = accum_time + exec_time
curr_fps = curr_fps + 1
if accum_time > 1:
accum_time = accum_time - 1
fps = "FPS: " + str(curr_fps)
curr_fps = 0
cv2.putText(result, str(number), (20, 40), cv2.FONT_HERSHEY_SIMPLEX, 1,
(0, 0, 255), 5)
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers)
# 画出跟踪轨迹
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(result, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr], 4)
cv2.imshow('Tracking', result)
cv2.namedWindow("result", cv2.WINDOW_NORMAL)
cv2.imshow("result", result)
#if isOutput:
# out.write(result)
if cv2.waitKey(100) & 0xFF == ord('q'):
break
yolo.close_session()
def task1_1(mogthr, inputpath, dataset):
# Create opencv video capture object
path = inputpath + 'in%06d.jpg'
cap = cv2.VideoCapture(path)
# Create Object Detector
detector = Detectors(thr=mogthr, dataset=dataset)
# Create Object Tracker
if dataset == 'highway':
tracker = Tracker(200, 0, 60, 100) # Tracker(200, 0, 200, 100)
elif dataset == 'traffic':
tracker = Tracker(200, 0, 60, 100) # Tracker(50, 0, 90, 100)
elif dataset == 'ownhighway':
tracker = Tracker(45, 0, 60, 100) # Tracker(50, 0, 90, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
if dataset == 'highway':
pts1 = np.float32([[120, 100], [257, 100], [25, 200], [250, 200]])
elif dataset == 'traffic':
pts1 = np.float32([[0, 50], [160, 15], [110, 190], [320, 110]])
elif dataset == 'ownhighway':
pts1 = np.float32([[190, 100], [290, 100], [60, 200], [250, 200]])
pts2 = np.float32([[0, 0], [320, 0], [0, 240], [320, 240]])
M = cv2.getPerspectiveTransform(pts1, pts2)
print M
counter = 0
# Infinite loop to process video frames
while True:
counter += 1
# Capture frame-by-frame
ret, frame = cap.read()
# Stop when no frame
if frame is None:
break
# Make copy of original frame
orig_frame = copy.copy(frame)
# Skip initial frames that display logo
#if (skip_frame_count < 200):
# skip_frame_count += 1
# continue
# Detect and return centeroids of the objects in the frame
centers, xd, yd, wd, hd = detector.Detect(frame, counter)
#print xd
vel = []
# If centroids are detected then track them
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers, dataset)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
vel = []
a=0
for j in range(5, len(tracker.tracks[i].trace) - 1):
a=a+1
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)), (int(x2), int(y2)), track_colors[clr], 2)
if dataset == 'highway':
if y1 > 100 and y2 < 200:
x1r, y1r, z1r = np.dot(M,[x1, y1, 1])
x2r, y2r, z2r = np.dot(M, [x2, y2, 1])
x1r, y1r = x1r/z1r, y1r/z1r
x2r, y2r = x2r / z2r, y2r / z2r
dist = np.float(
np.sqrt(((int(x2r) - int(x1r)) ** 2) + ((int(y2r) - int(y1r)) ** 2))) * np.float(
30) / 20 * np.float(24) / 5 # euclidean distance between two points
vel.append(dist)
if dataset == 'ownhighway':
if y1 > 100 and y2 < 200:
x1r, y1r, z1r = np.dot(M,[x1, y1, 1])
x2r, y2r, z2r = np.dot(M, [x2, y2, 1])
x1r, y1r = x1r/z1r, y1r/z1r
x2r, y2r = x2r / z2r, y2r / z2r
dist = np.float(
np.sqrt(((int(x2r) - int(x1r)) ** 2) + ((int(y2r) - int(y1r)) ** 2))) * np.float(
18) / 20 * np.float(24) / 5 # euclidean distance between two points
vel.append(dist)
if not vel == []:
#if i==1:
#print xd[i]#'value ' + xd[i] + ' value ' + yd[i]#+ ' frame '+frame+ ' vel ' +vel
# if dataset == 'ownhighway':
# #if i==0:
# print counter,i, xd,np.mean(vel)
# if counter>0:
# a=0
#
#
# #if xd==[]
# # if len(vel)<4: #and int(np.mean(vel))>100:
# # cv2.putText(frame, ' vel ' + str(int(np.mean(vel))), (int(xd[a]), int(yd[a] - 4)),
# # cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 1, cv2.LINE_AA)
# if len(vel)>3:# and int(np.mean(vel))>100:
# cv2.putText(frame, ' vel ' + str(int(np.mean(vel[-3:-1]))), (int(xd[0]), int(yd[0])),
# cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 1, cv2.LINE_AA)
# #cv2.putText(frame, ' vel ' + str(int(np.mean(vel))), (int(xd[0]), int(yd[0] - 4)),
# # cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 1, cv2.LINE_AA)
# #print int(np.mean(vel)),i,j
if dataset == 'ownhighway':
#print i, xd
if len(vel)<10:
cv2.putText(frame, ' vel ' + str(int(np.mean(vel))), (int(xd[0]), int(yd[0] - 4)),
cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 1, cv2.LINE_AA)
else:
cv2.putText(frame, ' vel ' + str(int(np.mean(vel[-10:-1]))), (int(xd[0]), int(yd[0] - 4)),
cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 1, cv2.LINE_AA)
if dataset == 'highway':
#print i, xd
if len(vel)<20:
cv2.putText(frame, ' vel ' + str(int(np.mean(vel))), (int(xd[i]), int(yd[i] - 4)),
cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 1, cv2.LINE_AA)
else:
cv2.putText(frame, ' vel ' + str(int(np.mean(vel[-20:-1]))), (int(xd[i]), int(yd[i] - 4)),
cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 1, cv2.LINE_AA)
#print int(np.mean(vel)), i, j
# x1 = tracker.tracks[i].trace[-2][0][0]
# y1 = tracker.tracks[i].trace[-2][1][0]
# x2 = tracker.tracks[i].trace[-1][0][0]
# y2 = tracker.tracks[i].trace[-1][1][0]
# if dataset == 'highway':
# if y1 > 100 and y2 < 200:
# x1r, y1r, z1r = np.dot(M,[x1, y1, 1])
# x2r, y2r, z2r = np.dot(M, [x2, y2, 1])
# x1r, y1r = x1r/z1r, y1r/z1r
# x2r, y2r = x2r / z2r, y2r / z2r
#
#
#
#
# dist = np.float(np.sqrt(((int(x2r) - int(x1r))**2) + ((int(y2r) - int(y1r))**2))) * np.float(30)/20 * np.float(24)/5#euclidean distance between two points
# vel.append(dist)
# cv2.putText(frame, ' vel '+str(int(dist)), (int(xd[i]), int(yd[i]-4)), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0,255,0),1,cv2.LINE_AA)
#print (x1, x2, y1, y2, dist, i)
# x1r,y1r = M * [x1,y1,1] #(x,y,1) = M * (xold,yold,1)
# x2r, y2r = M * [x2,y2,1]
#
# vel.append[j] = int(np.sqrt(((int(x2r) - int(x1r)) ** 2) + (
# (int(y2r) - int(y1r)) ** 2))) * int(3/50) * int(24/5) # * (m/pixel) * (frame/sec) # euclidean distance between two points
#
# if len(vel[j] > 10):
# return#velocity = np.mean(vel[j](i:i-10)) # * (m/pixel) * (frame/sec)
#print 'car '+ str(i) +' velocity '+ str(dist) #(x pixels every frame) -> * (m/pixel) * (frame/sec) = (m/sec)
# Display homography
dst = cv2.warpPerspective(frame, M, (320, 240))
cv2.imshow('Homography', dst)
cv2.imwrite('../week5/results/hom' + str(counter) + '.png', dst)
# Display the resulting tracking frame
cv2.imshow('Tracking', frame)
cv2.imwrite('../week5/results/out' + str(counter) + '.png', frame)
cv2.imwrite('out' + str(frame) + '.jpg',frame)
# Display the original frame
cv2.imshow('Original', orig_frame)
# Slower the FPS
cv2.waitKey(1)
# Check for key strokes
k = cv2.waitKey(1) & 0xff
if k == 27: # 'esc' key has been pressed, exit program.
break
if k == 112: # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if (pause is True):
print("Code is paused. Press 'p' to resume..")
while (pause is True):
# stay in this loop until
key = cv2.waitKey(30) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
def Trace_tracking(images):
"""tracking mutiple traces"""
#extract tfirst file name
first_filename = os.path.splitext(images[0])[0]
#extract first file index
#first_number = int(filter(str.isdigit, first_filename))
first_number = int(re.search(r'\d+', first_filename).group(0))
#define trace result path
outfile = save_path_track + str('{:04}'.format(first_number)) + '.txt'
print(outfile)
image_path = os.path.join(dir_path, images[0])
#load image from image path
frame = cv2.imread(image_path)
#Determine the width and height from the first image
height, width, channels = frame.shape
length = len(images)
print("Image sequence size: {0} {1} {2}\n".format(width, height, length))
#Create Object Detector
detector = Detectors()
#detector = Root_Detectors(pattern_id)
# Create Object Tracker, arguments:
# dist_thresh, max_frames_to_skip, max_trace_length, trackIdCount
#distance threshold. When exceeds the threshold, track will be deleted and new track is created
tracker = Tracker(dist_thresh, max_frames_to_skip, max_trace_length,
trackIdCount)
# Variables initialization
skip_frame_count = 0
#frame ID
ID = 0
#stem_track = np.zeros(3)
#initilize parameters for record radius and center locations
radius_track = []
centers_track = []
#Begin of process each image for tracking
###################################################################################
# loop to process video frames
for frame_ID, image in enumerate(images):
# Capture frame-by-frame
image_path = os.path.join(dir_path, image)
#load image frame
frame = cv2.imread(image_path)
# exit the loop if reach the end frame
if ID == len(images):
print("End of frame sequence!")
break
# Make copy of original frame
orig_frame = copy.copy(frame)
print("Processing frame {}...".format(frame_ID))
# Detect and return centeroids of the objects in the frame
(centers, radius_rec) = detector.Detect(frame, ID, radius_min,
radius_max)
# record radius and center locations
radius_track.append(radius_rec)
centers_track.append(centers)
#centers, stem_center = detector.Detect_root(frame, ID, pattern_id, stem_track)
#centers = detector.Detect_root_blob(frame, ID, pattern_id, stem_track)
# If centroids are detected then track them
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers)
print("Tracker size: {}...".format(len(tracker.tracks)))
#End of process each image for tracking
###################################################################################
radius_track = np.hstack(radius_track)
coord_radius = []
# combine x, y coordinates
for i in range(0, len(centers_track)):
for j in range(0, len(centers_track[i])):
coord_radius.append(np.array(centers_track[i][j]))
coord_radius = np.array(coord_radius)
#start index value along Z axis
offset = first_number
# write output as txt file
with open(outfile, 'w') as f:
#loop all tracked objects
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 2):
#accquire dimension of current tracker
dim = len(tracker.tracks[i].trace)
#extract point data from current tracker
point = np.asarray(tracker.tracks[i].trace)
#print(type(tracker.tracks[i].trace))
# accquire shape of points
nsamples, nx, ny = point.shape
#reshape points
point = point.reshape((nsamples, nx * ny))
#extract x,y,z coordinates
x = np.asarray(point[:, 0]).flatten()
y = np.asarray(point[:, 1]).flatten()
z = np.asarray(range(offset, dim + offset)).flatten()
#curve fitting of xy trace in 2D space
#popt, pcov = curve_fit(func, x, y)
#y = func(x, *popt)
#compute average radius
avg_radius = center_radius(x, y, radius_track, coord_radius)
#reshape radius array
r = np.asarray(avg_radius * np.ones((len(x), 1))).flatten()
#print("Average radius: {0} \n".format(avg_radius))
# write out tracing trace result
#if ( (len(x) == len(y) == len(z)) and (np.count_nonzero(x) == dim) and (np.count_nonzero(y) == dim) and sum(x) !=0 ):
if ((len(x) == len(y) == len(z)) and sum(x) != 0):
# save trace points as txt file
f.write("#Trace {0} \n".format(i))
np.savetxt(f, np.c_[x, y, z, r], fmt='%-7.2f')
#else:
#print("Inconsistant length pf 3D array!")
#ax.scatter(x, y, z, c = 'b', marker = 'o')
#ax.plot(x, y, z, label='Tracking root trace')
#f.write("#End\n")
# write end mark and close file
f.write("#End\n")
f.close()
def detect_video(yolo, video_path, output_path=""):
vid = cv2.VideoCapture(video_path)
tracker = Tracker(30, 0, 6, 0)
width = 400
height = 300
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127), (50, 50, 98), (37, 37, 47)]
pause = False
if not vid.isOpened():
raise IOError("Couldn't open webcam or video")
video_FourCC = int(vid.get(cv2.CAP_PROP_FOURCC))
video_fps = vid.get(cv2.CAP_PROP_FPS)
video_size = (int(vid.get(cv2.CAP_PROP_FRAME_WIDTH)),
int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT)))
isOutput = True if output_path != "" else False
if isOutput:
print("!!! TYPE:", type(output_path), type(video_FourCC),
type(video_fps), type(video_size))
out = cv2.VideoWriter(output_path, video_FourCC, video_fps,
(int(0.78 * width) - int(0.109 * width),
height - int(height * 0.0608)))
fps = "FPS: ??"
accum_time = 0
curr_fps = 0
clr = 0
frame_counter = 0
direction = ""
saving_name = ""
(dX, dY) = (0, 0)
lane_number = 0
centers = []
prev_time = timer()
while True:
return_value, frame = vid.read()
if return_value is False:
print('frame is empty: break')
break
orig_frame = frame.copy()
frame = cv2.resize(frame, (width, height))
frame = frame[int(0.0608 * height):height,
int(0.109 * width):int(0.78 * width)]
h, w, d = frame.shape
frame_counter += 1
image = Image.fromarray(frame)
new_result, centers, scores, out_classes, bbox = yolo.detect_image(
image)
result = np.array(new_result)
curr_time = timer()
exec_time = curr_time - prev_time
prev_time = curr_time
accum_time = accum_time + exec_time
curr_fps = curr_fps + 1
if len(centers) > 0 and len(out_classes) > 0:
tracker.Update(centers)
for i in range(len(tracker.tracks)):
x1 = 0
y1 = 0
x2 = 0
y2 = 0
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
y1 = tracker.tracks[i].trace[j][0][0]
x1 = tracker.tracks[i].trace[j][1][0]
y2 = tracker.tracks[i].trace[j + 1][0][0]
x2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id
try:
some_position = tracker.tracks[i].trace[2][1][0]
some_position_y = tracker.tracks[i].trace[2][0][0]
except:
some_position = tracker.tracks[i].trace[1][1][0]
some_position_y = tracker.tracks[i].trace[1][0][0]
#cv2.circle(result, (int(x2), int(y2)), 2, track_colors[clr%3], -1)
cv2.arrowedLine(
result, (int(some_position), int(some_position_y)),
(int(x2), int(y2)),
track_colors[clr % 3],
line_type=cv2.LINE_AA,
thickness=1)
dX = x2 - some_position
#dXCheck = x2-x1
dY = y2 - y1
(dirX, dirY) = ("", "")
#print(dX)
if np.abs(dX) >= 12:
dirX = "East" if np.sign(dX) == 1 else "West"
#if np.abs(dY)>=4:
# dirY = "North" if np.sign(dY) == 1 else "South"
if dirX != "" and dirY != "":
direction = "{}-{}".format(dirY, dirX)
else:
direction = dirX if dirX != "" else dirY
if direction == "East":
if (y2 > 0 and y2 <= h / 2.28):
lane_number = 1
if (y2 > h / 3.68 and y2 <= h / 1.35):
lane_number = 2
if (y2 > h / 1.35 and y2 <= h):
lane_number = 3
cv2.putText(result, 'WRONG DIRECTION',
(int(x1) - 25, int(y1) - 25),
cv2.FONT_HERSHEY_SIMPLEX, 0.30,
(0, 0, 255), 1)
now = dt.datetime.now()
raw_dir = './result/wrong_direction_'
lane_ = str(lane_number) + '_{0:02d}'.format(i)
seconds = now.second
right_now = now.strftime("%Y%m%d_%H%M%S_")
saving_name = "%s%s%s.jpg" % (raw_dir, right_now,
lane_)
if os.path.exists(saving_name) == False:
if os.path.exists("./result") == False:
os.mkdir("./result")
print(saving_name)
resized = resize(result)
cv2.imwrite(saving_name, resized)
else:
cv2.putText(result, direction,
(int(x1) - 25, int(y1) - 25),
cv2.FONT_HERSHEY_SIMPLEX, 0.30,
(0, 255, 0), 1)
if accum_time > 1:
accum_time = accum_time - 1
fps = "FPS: " + str(curr_fps)
curr_fps = 0
cv2.putText(result, "Count: " + str(len(centers)), (3, 35),
cv2.FONT_HERSHEY_SIMPLEX, 0.4, (0, 255, 0), 1)
cv2.putText(result,
text=fps,
org=(3, 15),
fontFace=cv2.FONT_HERSHEY_SIMPLEX,
fontScale=0.40,
color=(0, 255, 0),
thickness=1)
cv2.namedWindow("result", cv2.WINDOW_NORMAL)
cv2.imshow("result", result)
k = cv2.waitKey(1) & 0xff
if isOutput:
try:
out.write(result)
except:
continue
if k == ord('q'):
break
if k == 112: # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if (pause is True):
print("Code is paused. Press 'p' to resume..")
while (pause is True):
# stay in this loop until
key = cv2.waitKey(30) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
yolo.close_session()
th1 = cv2.dilate(th1, kernel, iterations=4)
th1 = cv2.erode(th1, kernel, iterations=2)
cv2.imshow('diff', th1)
t1 = time.time()
find_frame, our_team_point, enemy_team_point, other_point = gop.main(
frame)
# If centroids are detected then track them
if (len(our_team_point) > 0):
KF_our_team = []
# Track object using Kalman Filter
tracker.Update(our_team_point)
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
KF_our_team.append([x2, y2])
# If centroids are detected then track them
if (len(enemy_team_point) > 0):
KF_enemy_team = []
# Track object using Kalman Filter
tracker.Update(enemy_team_point)
def main():
# Create opencv video capture object
cap = cv2.VideoCapture('output.avi')
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(160, 30, 5, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
while cap.isOpened():
# Capture frame-by-frame
ret, frame = cap.read()
# Make copy of original frame
orig_frame = copy.copy(frame)
# Skip initial frames that display logo
if skip_frame_count < 15:
skip_frame_count += 1
continue
# Detect and return centeroids of the objects in the frame
centers = detector.Detect(frame)
# If centroids are detected then track them
if len(centers) > 0:
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if len(tracker.tracks[i].trace) > 1:
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 2)
# Display the resulting tracking frame
cv2.imshow('Tracking', frame)
# Display the original frame
cv2.imshow('Original', orig_frame)
# Slower the FPS
cv2.waitKey(50)
# Check for key
k = cv2.waitKey(50) & 0xff
if k == 27: # 'esc' key has been pressed, exit program.
break
if k == 112: # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if pause is True:
print("Code is paused. Press 'p' to resume..")
while pause is True:
# stay in this loop until
key = cv2.waitKey(30) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
def main():
"""Main function for multi object tracking
Usage:
$ python2.7 objectTracking.py
Pre-requisite:
- Python2.7
- Numpy
- SciPy
- Opencv 3.0 for Python
Args:
None
Return:
None
"""
'''
options = {
'model': 'cfg/tiny-yolo-voc-1c.cfg',
'load': 4000,
'threshold': 0.15,
'gpu': 1.0
}
tfnet = TFNet(options)
'''
net = Detector(bytes("cfg/yolov3.cfg", encoding="utf-8"),
bytes("weights/yolov3.weights", encoding="utf-8"), 0,
bytes("cfg/coco.data", encoding="utf-8"))
# Create opencv video capture object
cap = cv2.VideoCapture('mansiroad_trimmed.mp4')
#length = cap.get(cv2.CAP_PROP_FRAME_COUNT)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, 1920)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 1080)
'''
# Create opencv video capture object
cap = cv2.VideoCapture('data/TrackingBugs.mp4')
# Create Object Detectorx
detector = Detectors()
'''
# Create Object Tracker
tracker = Tracker(25, 60, 1000, 10)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
frame_array = []
first = 0
# Infinite loop to process video frames
size = (1920, 1080)
c = 0
#print(length)
#while(c<=length-1):
try:
while True:
# Capture frame-by-frame
ret, frame = cap.read()
#print('ret',ret)
# Make copy of original frame
orig_frame = copy.copy(frame)
# Skip initial frames that display logo
'''
if (skip_frame_count < 15):
skip_frame_count += 1
continue
'''
#print('ssss')
# Detect and return centeroids of the objects in the frame
centers = vid.detect(ret, frame, net) #,tfnet)
print("centers :", centers)
# If centroids are detected then track them
if (len(centers) > 0):
first = 1
# Track object using Kalman Filter
tracker.Update(centers, first)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
print('NUM OF OBJECTS : ', len(tracker.tracks))
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
#print('NUM OF OBJECTS : ',tracker.tracks[i].trace)
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr], 2)
# Display the resulting tracking frame
#cv2.imshow('Tracking', frame)
elif first == 1:
tracker.Update(centers, 0)
print('NUM OF OBJECTSno : ', len(tracker.tracks))
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
print(
'NUM OF OBJECTSnononono : ',
len(tracker.tracks[i].trace),
)
print(
'trace : ',
tracker.tracks[i].trace[
len(tracker.tracks[i].trace) - 1],
)
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr], 2)
height, width, layers = frame.shape
size = (width, height)
frame_array.append(frame)
cv2.imshow('ss', frame)
'''
# Display the original frame
#cv2.imshow('Original', orig_frame)
if keyboard.is_pressed('q'):# 'q' key has been pressed, exit program.
break
# Slower the FPS
cv2.waitKey(50)
# Check for key strokes
k = cv2.waitKey(50) & 0xff
if k == 27: # 'esc' key has been pressed, exit program.
break
if k == 112: # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if (pause is True):
print("Code is paused. Press 'p' to resume..")
while (pause is True):
# stay in this loop until
key = cv2.waitKey(30) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
'''
key = cv2.waitKey(1) & 0xFF
# Exit
if key == ord('q'):
break
# Take screenshot
if key == ord('s'):
cv2.imwrite('frame_{}.jpg'.format(time.time()), frame)
c += 1
except:
print('Video Ended')
# When everything done, release the capture
finally:
#out = cv2.VideoWriter('result2.mp4',cv2.VideoWriter_fourcc(*'MP4V'), int(cap.get(cv2.CAP_PROP_FPS)), size)
#print('11')
#for i in range(len(frame_array)):
# writing to a image array
#cv2.imshow('ff',frame_array[i])
# writing to a image array
#out.write(frame_array[i])
#out.release()
#out.release()
cap.release()
cv2.destroyAllWindows()
def main(tiffStep):
"""Main function for multi object tracking
"""
#tiffStep = 512
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(200, 50, 25, 100)
# Variables initialization
pause = False
track_colors = [random_color() for x in xrange(256)]
# Infinite loop to process video frames
stTmAv = time.time()
outFName = imFolder + '_traced_0-' + str(tiffStep) + '.tiff'
#memmap_image = tifffile.memmap(outFName, shape=(tiffStep, newx, newy, 3), dtype='uint8')
imgs = np.zeros((tiffStep, newy, newx, 3), dtype=np.uint8)
tTm = 0
stTm = time.time()
for fr in xrange(len(flyContours[0])):
# Capture frame-by-frame
frame = getBgSubIm((flyContours[0][fr], flyContours[1]))
frame = cv2.cvtColor(frame, cv2.COLOR_GRAY2BGR)
outFrame = cv2.cvtColor(flyContours[0][fr], cv2.COLOR_GRAY2BGR)
# Detect and return centeroids of the objects in the frame
centers = detector.DetectCM(frame)
# If centroids are detected then track them
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id
cv2.line(outFrame, (int(x1), int(y1)),
(int(x2), int(y2)), track_colors[clr], 2)
cv2.circle(outFrame, (int(x2), int(y2)), 2,
track_colors[clr], 2)
#cv2.circle(outFrame, (int(x1), int(y1)), 2, (255,25,255), 2)
cv2.circle(outFrame, (int(x2), int(y2)), 2,
track_colors[clr], 1)
## Display the resulting tracking frame
#cv2.imshow('Tracking', outFrame)
#cv2.waitKey(1)
# outFName = imFolder+'_traced/'+flist[fr].split('/')[-1]
# cv2.imwrite(outFName, outFrame)
img = cv2.resize(outFrame, (newx, newy))
imN = (fr % tiffStep)
if (imN == 0 and fr > 0):
outFName = imFolder + '_traced_' + str(
fr - tiffStep) + '-' + str(fr) + '.tiff'
startNT(imageio.mimwrite, (outFName, imgs))
imgs = np.zeros((tiffStep, newy, newx, 3), dtype=np.uint8)
#memmap_image = tifffile.memmap(outFName, shape=(tiffStep, newx, newy, 3), dtype='uint8')
#memmap_image[imN] = img
tm = time.time()
fps = (tiffStep / (tm - stTm))
tTm += tm - stTm
print('FPS: %0.3f (frame# %d)' % (fps, fr))
stTm = tm
#else:
# #print fr, imN
imgs[imN] = img
imageio.mimwrite(
imFolder + '_traced_' + str(
(fr / tiffStep) * tiffStep) + '-' + str(fr) + '.tiff', imgs[:imN])
print('Tracking average FPS: %0.3f' % (float(fr) / (time.time() - stTmAv))
) #(1.0/(tm-stTm)))
cv2.destroyAllWindows()
def main():
parser = argparse.ArgumentParser(description='Description of your program')
parser.add_argument('-vid',
'--video',
required=True,
default="Input/project.avi",
help="Video File Path")
parser.add_argument('-roi',
'--roi creation mode',
required=False,
default="manually",
help="Create region of interest-do it 'manually'," +
"or use the 'pre-tested' one which gives good results")
args = vars(parser.parse_args())
video = args['video']
roi_mode = args['roi creation mode']
videopath, __ = video.split(".", -1)
__, videoname = videopath.split('/', -1)
print("video", video)
print("videopath", videopath)
print("videoname", videoname)
camera = cv2.VideoCapture(video)
ret, frame = camera.read()
human_detect = Human_Detectors()
# Initialise Tracker
tracker = Tracker(160, 30, 5, 100)
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
if roi_mode == 'manually':
roi = cv2.selectROI(frame, showCrosshair=False)
elif roi_mode == 'pre-tested':
try:
roi_file = open(videopath + '_pre-testedROI.txt', 'r')
rois = roi_file.read()
rois = rois[1:-1]
roi = rois.split(", ")
for i in range(len(roi)):
roi[i] = int(roi[i])
except:
print(
"The pre-tested Region of Interest file does not exist yet. Please create it manually."
)
roi = cv2.selectROI(frame, showCrosshair=False)
cv2.destroyWindow('ROI selector')
fourcc = cv2.VideoWriter_fourcc(*'MJPG')
vid_out = cv2.VideoWriter('Output_Video/' + videoname + '.avi',
fourcc,
20.0, (int(camera.get(3)), int(camera.get(4))),
isColor=True)
# Run through video frames
while (camera.isOpened()):
ret, frame = camera.read()
# Detect and return centeroids of the objects in the frame
centers = human_detect.Detect(frame)
total_centers = len(centers)
# Track centroids, if found
if (total_centers > 0):
# Track object using Kalman Filter
tracker.Update(centers)
count = len(tracker.tracks)
# drawing tracks with different colors
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Coordinated of predicted line
x_1 = int(tracker.tracks[i].trace[j][0][0])
y_1 = int(tracker.tracks[i].trace[j][1][0])
x_2 = int(tracker.tracks[i].trace[j + 1][0][0])
y_2 = int(tracker.tracks[i].trace[j + 1][1][0])
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (x_1, y_1), (x_2, y_2),
track_colors[clr], 2)
# Show the tracked video frame
vid_out.write(frame)
cv2.rectangle(frame, (roi[0], roi[1]),
(roi[0] + roi[2], roi[1] + roi[3]), (0, 255, 0), 1)
cv2.putText(frame, 'people detected: ' + str(count), (10, 450),
cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 1,
cv2.LINE_AA)
cv2.imshow('Tracking', frame)
#cv2.imshow('rectangle', rect)
print("total number people in the frame: ", count)
key = cv2.waitKey(50) & 0xff
# Escape key to exit
if key == 27:
break
camera.release()
vid_out.release()
cv2.destroyAllWindows()
if roi_mode == 'manually':
print(
"Do you wish to save the created roi to be used as an optional pre-tested file for this video next run onwards (if it gave good results)"
)
wr = create_roi(videoname, roi)
def main():
"""Main function for multi object tracking
Usage:
$ python2.7 objectTracking.py
Pre-requisite:
- Python2.7
- Numpy
- SciPy
- Opencv 3.0 for Python
Args:
None
Return:
None
"""
# Create opencv video capture object
cap = cv2.VideoCapture('data/TrackingBugs.mp4')
"""
Resolution: 596 x 336
Frame rate: 30 fps
"""
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(30, 30, 10, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
# Infinite loop to process video frames
while (True):
# Capture frame-by-frame
ret, frame = cap.read()
# Make copy of original frame
orig_frame = copy.copy(frame)
# Skip initial frames that display logo
if (skip_frame_count < 15):
skip_frame_count += 1
continue
# Detect and return centeroids of the objects in the frame
centers = detector.Detect(frame)
# If centroids are detected then track them
if (len(centers) > 0):
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
position = tracker.tracks[i].position()
error = tracker.tracks[i].position_error()
cv2.circle(frame, (position[0], position[1]), error,
(200, 200, 200), 2)
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 2)
# Display the resulting tracking frame
cv2.imshow('Tracking', frame)
# Display the original frame
cv2.imshow('Original', orig_frame)
# Slower the FPS
cv2.waitKey(1)
# Check for key strokes
k = cv2.waitKey(50) & 0xff
if k == 27: # 'esc' key has been pressed, exit program.
break
if k == 112: # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if (pause is True):
print("Code is paused. Press 'p' to resume..")
while (pause is True):
# stay in this loop until
key = cv2.waitKey(30) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
# If centroids are detected then track them
if (len(rectList) > 0):
centers = [None] * len(rectList)
for i in range(len(rectList)):
rect = rectList[i]
x = int(rect[0] + rect[2] / 2)
y = int(rect[1] + rect[3] / 2)
b = np.array([[x], [y]])
centers[i] = np.round(b)
cv2.rectangle(frame, rectList[i], (0, 0, 255), 5)
cv2.circle(frame, (x, y), 4, (0, 0, 255), 3)
# Track object using Kalman Filter
if skip_frame_count > 10:
tracker.Update(centers, False)
skip_frame_count = skip_frame_count + 1
if debug:
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].id % 9