def DispImg(self):
img = cv2.cvtColor(self.Image, cv2.COLOR_BGR2RGB)
results = dn.detect_np(self.net, self.meta, img)
img = dn.chinese_to_img(self.font, img, results)
qimg = qimage2ndarray.array2qimage(img)
self.DispLb.setPixmap(QPixmap(qimg))
self.DispLb.show()
def detect_object(self,event):
ret, frame = self.cap.read()
k = cv2.waitKey(1)
r = detect_np(self.net, self.meta, frame)
result_im = check_detected_result_image(frame, r)
cv2.imshow('Image', result_im )
cv2.waitKey(1)
#check obstacle
flg = False
for indx in range(len(r)): #Check detected objects
for indx2 in range(len(self.obstacle_label)): #check label name
#Check Object Label
if r[indx][0] == self.obstacle_label[indx2]:
ratio = r[indx][2][2]*r[indx][2][3]/self.imsize
if ratio > self.obstacle_ratio: #check object size (ratio per frame size)
flg = True
break
if flg == True:
break
if flg == True:
self.obstacle_cnt = min(self.obstacle_cnt + 1,self.obstacle_cancel_frame_th)
else:
self.obstacle_cnt = max(self.obstacle_cnt - 1, 0)
if self.obstacle_cnt > self.obstacle_frame_th:
self.obstacle_flg = True
elif self.obstacle_cnt <= 0:
self.obstacle_flg = False
self.object_detect_publisher.publish(self.obstacle_flg)
print(self.obstacle_cnt)
def detect(net,meta,cam_param,im,isshow=False,issave=False,thresh=.3, hier_thresh=.5, nms=.45,
name=None, output_dir=None):
print "detection ============================="
start_t = time.time()
r = dn.detect_np(net,meta,im)
bdict = {}
bdict = allposition(r,cam_param)
end_t = time.time()
print name, "detection result:",bdict
if isshow or issave :
image = im.copy()
for i in r:
# print i
sx1,sy1,sx2,sy2=i[2]
sx1=sx1-sx2/2
sy1=sy1-sy2/2
cv2.rectangle(image,(int(sx1),int(sy1)),(int(sx1+sx2),int(sy1+sy2)),(0,255,0),3)
if (sy1 > 10):
cv2.putText(image, i[0]+"-"+str(round(i[1],2)), (int(sx1),int(sy1-6)), cv2.FONT_HERSHEY_COMPLEX_SMALL,0.8, (0, 255, 0) )
else:
cv2.putText(image, i[0]+"-"+str(round(i[1],2)), (int(sx1),int(sy1+15)), cv2.FONT_HERSHEY_COMPLEX_SMALL,0.8, (0, 255, 0) )
if isshow:
cv2.imshow("pre",image)
cv2.waitKey(1)
# cv2.imwrite("current.jpg",image)
if issave:
if name is None:
name = 'test'
if output_dir is None:
output_dir = "./run_pic"
t = str(end_t-start_t)
cv2.imwrite(os.path.join(output_dir,"pre-"+name+"-"+t+".jpg"),image)
return bdict
def detect(cfg_file, weight_file, video_file):
dn.set_gpu(0)
net = dn.load_net(cfg_file, weight_file, 0)
meta = dn.load_meta("cfg/coco.data")
print("video: %s" % video_file)
cap = cv2.VideoCapture(video_file)
video_len = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
fps = int(cap.get(cv2.CAP_PROP_FPS))
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
terminate, is_paused = False, False
idx = 0
fourcc = cv2.VideoWriter_fourcc(*"mp4v")
video_writer = cv2.VideoWriter("gray_processed.mp4", fourcc, fps, (width, height))
while not terminate and cap.isOpened():
idx += 1
print("processing: %.4f%%" % (idx * 100.0 / video_len))
#print("debug")
if not is_paused:
ret, frame = cap.read()
if not ret:
break
boxes = dn.detect_np(net, meta, frame)
new_frame = my_plot(frame.copy(), boxes)
video_writer.write(new_frame)
"""
cv2.imshow('image', new_frame)
key = cv2.waitKey(1)
if key & 255 == 27: # ESC
print("terminating")
terminate = True
elif key & 255 == 32: # ' '
print("toggeling pause: " + str(not is_paused))
is_paused = not is_paused
elif key & 255 == 115: # 's'
print("stepping")
ret, frame = cap.read()
if not ret:
break
boxes = dn.detect_np(net, meta, frame)
is_paused = True
"""
cap.release()
video_writer.release()
def callback(ch, method, properties, body):
mdata = pickle.loads(body)
print('working on count:', mdata['count'])
#time.sleep(20)
result = darknet.detect_np(net, meta, mdata['buff'])
print(mdata['count'], result)
dic = {'bear': 0, 'zebra': 0}
for ob in result:
ani = ob[0].decode()
if ani in dic and dic[ani] < ob[1] * 100:
dic[ani] = int(ob[1] * 100)
sqlhead = ("INSERT INTO anitag "
"(BEAR, ZEBRA) "
"VALUES (%(bear)s, %(zebra)s)")
print(dic)
cur.execute(sqlhead, dic)
cnx.commit()
def detect_targets(self, frame):
'''
Detects targets in a frame using Darknet. Returns a list of detections
and appends detections to a history of detections
frame: The 2D image to run detections on.
'''
frame_size = (frame.shape[1], frame.shape[0])
frame = cv2.resize(frame, self.net_size)
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
# For use in debug. If use_temporal is "fake", appends black masks to the
# 4th and 5th channels. Otherwise, if use_temporal is True, appends masks
# from past frames to the image before passing it through the darknet detector.
if self.use_temporal == "fake":
augmented_frame = np.stack(
(frame[:, :, 0], frame[:, :, 1], frame[:, :, 2],
np.zeros((frame.shape[0], frame.shape[1]), dtype="uint8"),
np.zeros((frame.shape[0], frame.shape[1]), dtype="uint8")),
axis=-1)
elif self.use_temporal:
augmented_frame = np.stack(
(frame[:, :, 0], frame[:, :, 1], frame[:, :, 2],
self.make_dets_frame_mask('red', frame.shape),
self.make_dets_frame_mask('blue', frame.shape)),
axis=-1)
#cv2.imshow("Red", augmented_frame[:,:,3])
#cv2.imshow("Blue", augmented_frame[:,:,4])
result = dn.detect_np(self.net, self.meta,
augmented_frame if self.use_temporal else frame)
self.last_dets.insert(0, [
DetectedTarget(result_tuple, self.net_size, frame_size)
for result_tuple in result
])
if len(self.last_dets) > self.PAST_FRAME_NUM:
self.last_dets.pop()
return self.last_dets[0]
def callback(ch, method, properties, body):
'''
Read the frames queued in rabbitmq,
run darknet to indentify animals,
and save results to mysql databases.
'''
mdata = pickle.loads(body)
result = darknet.detect_np(net, meta, mdata['buff'])
# find the animals with max confid for each animal type in one frame
dic = {}
for ob in result:
ani = ob[0].decode()
if ani in aniset: dic[ani] = max(int(ob[1] * 100), dic.get(ani, 0))
if not dic: return
# save animal type, timestamp and confid into mysql db
for i in dic:
cur.execute("INSERT INTO anitag "
"(object,timestamp,confid)"
"VALUES ('{0}',{1},{2})".format(i, mdata['timestamp'],
dic[i]))
cnx.commit()
def detect(net,meta,cam_param,im,isshow=True,issave=False,thresh=.3, hier_thresh=.5, nms=.45, name=None):
r = dn.detect_np(net,meta,im)
bdict = {}
bdict = allposition(r,cam_param)
if isshow or issave :
for i in r:
# print i
sx1,sy1,sx2,sy2=i[2]
sx1=sx1-sx2/2
sy1=sy1-sy2/2
cv2.rectangle(im,(int(sx1),int(sy1)),(int(sx1+sx2),int(sy1+sy2)),(0,255,0),3)
if (sy1 > 10):
cv2.putText(im, i[0]+"-"+str(round(i[1],2)), (int(sx1),int(sy1-6)), cv2.FONT_HERSHEY_COMPLEX_SMALL,0.8, (0, 255, 0) )
else:
cv2.putText(im, i[0]+"-"+str(round(i[1],2)), (int(sx1),int(sy1+15)), cv2.FONT_HERSHEY_COMPLEX_SMALL,0.8, (0, 255, 0) )
if isshow:
cv2.imshow("pre",im)
cv2.waitKey(1)
if issave:
if name is None:
name = str(time.time())
cv2.imwrite("pre-"+name+".jpg",im)
return bdict
# net = Detector(bytes("cfg/yolov3-spp.cfg", encoding="utf-8"), bytes("cfg/yolov3-spp.weights", encoding="utf-8"), 0, bytes("cfg/coco.data",encoding="utf-8"))
img_name = args.img_name.encode("utf-8")
print('read img_name:', args.img_name)
img = cv2.imread(args.img_name)
# img2 = Image(img)
# img2 = darknet.Image(img2)
# r = net.classify(img2)
start_time = time.time()
# results = net.detect(img2,thresh=.2, hier_thresh=.2, nms=.2)
results = darknet.detect_np(net,
meta,
img,
thresh=.3,
hier_thresh=.3,
nms=.45)
end_time = time.time()
print('\tElapsed Time:{:.3f} sec'.format(end_time - start_time))
min_car_height = 90
frontcars = []
for i, obj in enumerate(results):
cat, score, bounds = obj
x, y, w, h = bounds
cv2.rectangle(img, (int(x - w / 2), int(y - h / 2)),
(int(x + w / 2), int(y + h / 2)), (255, 0, 0),
w = box[2]
h = box[3]
left = float(x)-0.5*float(w)
top = float(y)-0.5*float(h)
right = float(x)+0.5*float(w)
bot = float(y) + 0.5*float(h)
return (left,top,right,bot,mess,confidence)#(left,top,right,bot)
# Darknet
####################### for python3 ##########################
net = dn.load_net("detector2/yolov3.cfg".encode("ascii"), "detector2/yolov3.weights".encode("ascii"), 0)
meta = dn.load_meta("detector2/coco.data".encode("ascii"))
cv2.namedWindow('YOLOV3')
img = cv2.imread('dog.jpg')
d2_boxes = dn.detect_np(net, meta, img, 0.55, 0.55, 0.45)# net, meta, image, thresh, hier_thresh, nms
for b in d2_boxes:
boxResults = process_d2box(b)
if boxResults is None:
continue
left, top, right, bot, mess, confidence = boxResults
cv2.rectangle(img, (int(left), int(top)), (int(right), int(bot)),
(255, 0, 0), 2)
cv2.putText(img, str(mess) + ":" + str('%.1f' % (float(confidence) * 100)) + "%", (int(left), int(top) - 8),
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 1) # hy change:2017/12/18
cv2.imshow('YOLOV3',img)
cv2.waitKey()
cv2.destroyAllWindows()
def run(self):
self.net = dn.load_net(self.cfgFile, self.weightFile, 0)
self.meta = dn.load_meta(self.dataFile)
def nothing(t):
pass
cv2.namedWindow("Vision")
dataFile = open(self.dataFile, "r")
line = dataFile.readline()
namesFile = None
while line:
line = line.replace(" ", "")
line = line.split("=")
if line[0] == "names":
print "Names file:", line
namesFile = open(line[1].replace("\n", ""), "r")
break
line = dataFile.readline()
dataFile.close()
if namesFile != None:
line = namesFile.readline()
index = 0
while line:
self.detectionDictionary[line.replace("\n", "")] = index
index += 1
line = namesFile.readline()
namesFile.close()
if self.useVideo == False and self.usePicture == False:
bus = pc2.BusManager()
self.activeCamera = pc2.Camera()
self.activeCamera.connect(bus.getCameraFromIndex(0))
self.activeCamera.startCapture()
elif self.useVideo == True:
self.activeCamera = cv2.VideoCapture(self.videoPath)
elif self.usePicture == True:
self.sourceImg = cv2.imread(self.picturePath)
while self.running == True:
if self.useVideo == False and self.usePicture == False:
self.image = self.activeCamera.retrieveBuffer()
self.image = self.image.convert(pc2.PIXEL_FORMAT.BGR)
img = np.array(self.image.getData(), dtype="uint8").reshape(
(self.image.getRows(), self.image.getCols(), 3))
img = cv2.flip(img, -1)
elif self.useVideo == True:
t, img = self.activeCamera.read()
else:
img = copy.copy(self.sourceImg)
#yoloImage = dn.IMAGE()
w, h, c = img.shape
detections = dn.detect_np(self.net,
self.meta,
img,
thresh=.1,
hier_thresh=.1)
newDetections = []
self.imgPoints = []
self.srcPoints = []
for detection in detections:
fixedClassNumber = self.detectionDictionary[detection[0]]
newDetections.append(
[fixedClassNumber, detection[1], detection[2]])
if detection[0] in self.torpedoDictionary or detection[
0] == "Corner":
if detection[0] != "Corner":
self.imgPoints.append(
(detection[2][0], detection[2][1]))
self.srcPoints.append(self.torpedoSrcPoints[
self.torpedoDictionary[detection[0]]])
print "Detection", detection[0]
if detection[0] == "Top Right Hole":
pixelError = 20
x, y, width, height = detection[2]
x1, y1, x2, y2 = x + (
.5 * width) + pixelError, y + (
.5 * height) + pixelError, x + (
.5 * width) + pixelError, y + (
.5 * height) + pixelError,
if x1 < 0:
x1 = 0
if y1 < 0:
y1 = 0
if x2 >= w:
x2 = w - 1
if y2 >= h:
y2 = h - 1
print "Sub img", x1, y1, x2, y2
subImg = img[int(y1):int(y2), int(x1):int(x2)]
while True:
cv2.imshow("Sub", subImg)
if cv2.waitKey(1) == ord("q"):
break
elif detection[0] == "Corner" and False:
print "Found a corner"
pixelError = 20
'''cv2.circle(img, (int(detection[2][0]), int(detection[2][1])), 10, (255,0,0), -1)
while True:
cv2.imshow("Vision", img)
if cv2.waitKey(1) == ord(" "):
break'''
for secondDet in detections:
'''cv2.rectangle(img, (int(secondDet[2][0] - (.5 * secondDet[2][2]) - pixelError ),int(secondDet[2][1] - (.5 * secondDet[2][3]) - pixelError)), (int(secondDet[2][0] - (.5 * secondDet[2][2]) + pixelError),int(secondDet[2][1] - (.5 * secondDet[2][3]) + pixelError)), (255, 0, 0), 3)
cv2.rectangle(img, (int(secondDet[2][0]-(.5 * secondDet[2][2])), int(secondDet[2][1]- (.5 * secondDet[2][3]))), (int(secondDet[2][0] + (.5*secondDet[2][2])), int(secondDet[2][1] + (.5*secondDet[2][3]))), (0,0,255))
while True:
cv2.imshow("Vision", img)
if cv2.waitKey(1) == ord(" "):
break'''
if secondDet[0] in self.torpedoDictionary:
index = None
if detection[2][0] >= secondDet[2][0] - (
.5 * secondDet[2][2]
) - pixelError and detection[2][
0] <= secondDet[2][0] - (
.5 * secondDet[2][2]) + pixelError:
if detection[2][1] <= secondDet[2][1] - (
.5 * secondDet[2][3]
) + pixelError and detection[2][
1] > secondDet[2][1] - (
.5 *
secondDet[2][3]) - pixelError:
print "In top left corner of", secondDet[
0]
cv2.circle(img, (int(detection[2][0]),
int(detection[2][1])),
10, (255, 0, 0), -1)
index = self.torpedoDictionary[
secondDet[0]]
index += 6
elif detection[2][0] >= secondDet[2][0] + (
.5 * secondDet[2][2]
) - pixelError and detection[2][
0] <= secondDet[2][0] + (
.5 * secondDet[2][2]) + pixelError:
if detection[2][1] <= secondDet[2][1] - (
.5 * secondDet[2][3]
) + pixelError and detection[2][
1] > secondDet[2][1] - (
.5 *
secondDet[2][3]) - pixelError:
print "In top right corner of", secondDet[
0]
cv2.circle(img, (int(detection[2][0]),
int(detection[2][1])),
10, (0, 255, 0), -1)
index = self.torpedoDictionary[
secondDet[0]]
index += 3
if detection[2][0] >= secondDet[2][0] - (
.5 * secondDet[2][2]
) - pixelError and detection[2][
0] <= secondDet[2][0] - (
.5 * secondDet[2][2]) + pixelError:
if detection[2][1] <= secondDet[2][1] + (
.5 * secondDet[2][3]
) + pixelError and detection[2][
1] > secondDet[2][1] + (
.5 *
secondDet[2][3]) - pixelError:
print "In bottom left corner of", secondDet[
0]
cv2.circle(img, (int(detection[2][0]),
int(detection[2][1])),
10, (0, 0, 255), -1)
index = self.torpedoDictionary[
secondDet[0]]
index += 5
elif detection[2][0] >= secondDet[2][0] + (
.5 * secondDet[2][2]
) - pixelError and detection[2][
0] <= secondDet[2][0] + (
.5 * secondDet[2][2]) + pixelError:
if detection[2][1] <= secondDet[2][1] + (
.5 * secondDet[2][3]
) + pixelError and detection[2][
1] > secondDet[2][1] + (
.5 *
secondDet[2][3]) - pixelError:
print "In bottom right corner of", secondDet[
0]
cv2.circle(img, (int(detection[2][0]),
int(detection[2][1])),
10, (255, 0, 255), -1)
index = self.torpedoDictionary[
secondDet[0]]
index += 4
if index == None:
pass
else:
index += 3
self.srcPoints.append(
self.torpedoSrcPoints[index])
self.imgPoints.append(
(int(detection[2][0]),
int(detection[2][1])))
if len(self.imgPoints) >= 4:
print "Solving..."
#print self.imgPoints, self.srcPoints
rvec, tvec = cv2.solvePnP(
np.array(self.srcPoints).astype("float32"),
np.array(self.imgPoints).astype("float32"),
np.array(self.camMat).astype("float32"), np.zeros(
(4, 1)))[-2:]
(pose1, jacobian) = cv2.projectPoints(
np.array([(0.0, 0.0, 0.1)]), rvec, tvec,
np.array(self.camMat).astype("float32"), None)
(pose, jacobian) = cv2.projectPoints(
np.array([(0, 0, 12.0)]), rvec, tvec,
np.array(self.camMat).astype("float32"), None)
cv2.line(img, (int(pose1[0][0][0]), int(pose1[0][0][1])),
(int(pose[0][0][0]), int(pose[0][0][1])), (255, 0, 0),
2)
print "Rvec:", rvec, "\n"
print "Tvec:", tvec
s = time.time()
while time.time() - s < 0:
cv2.imshow("Vision", img)
k = cv2.waitKey(1)
if k == ord(" "):
break
for detection in detections:
loc = detection[2]
cv2.rectangle(
img,
(int(loc[0] - (.5 * loc[2])), int(loc[1] - (.5 * loc[3]))),
(int(loc[0] + (.5 * loc[2])), int(loc[1] + (.5 * loc[3]))),
(0, 0, 255))
self.getList.append(newDetections)
cv2.imshow("Vision", img)
key = cv2.waitKey(1)
if key == ord('q'):
self.running = False
self.activeCamera.stopCapture()
def run(self):
if not onLinux:
return
self.net = dn.load_net(self.cfgFile, self.weightFile, 0)
self.meta = dn.load_meta(self.dataFile)
dataFile = open(self.dataFile, "r")
line = dataFile.readline()
namesFile = None
while line:
line = line.replace(" ", "")
line = line.split("=")
if line[0] == "names":
print "Names file:", line
namesFile = open(line[1].replace("\n", ""), "r")
break
line = dataFile.readline()
dataFile.close()
if namesFile != None:
line = namesFile.readline()
index = 0
while line:
self.detectionDictionary[line.replace("\n", "")] = index
index += 1
line = namesFile.readline()
namesFile.close()
if useVideo == True:
video = cv2.VideoCapture(videoFile)
else:
bus = pc2.BusManager()
self.activeCamera = pc2.Camera()
self.frontCamera = pc2.Camera()
self.botCamera = pc2.Camera()
self.frontCamera.connect(bus.getCameraFromIndex(1))
self.frontCamera.startCapture()
time.sleep(1)
self.botCamera.connect(bus.getCameraFromIndex(0))
self.botCamera.startCapture()
self.activeCamera = self.frontCamera
while self.running == True:
if self.useVideo == False:
self.image = self.activeCamera.retrieveBuffer()
self.image = self.image.convert(pc2.PIXEL_FORMAT.BGR)
img = np.array(self.image.getData(), dtype="uint8").reshape((self.image.getRows(), self.image.getCols(), 3))
img = cv2.flip(img, -1)
else:
t, img = self.activeCamera.read()
#yoloImage = dn.IMAGE()
detections = dn.detect_np(self.net, self.meta, img, thresh=.1, hier_thresh = .1)
newDetections = []
self.imgPoints = []
self.srcPoints = []
for detection in detections:
fixedClassNumber = self.detectionDictionary[detection[0]]
cv2.putText(img, detection[0], (int(detection[2][0] - (.5 * detection[2][2])), int(detection[2][1] - (.5 * detection[2][3]))), cv2.FONT_HERSHEY_SIMPLEX, .5, (0,0,255), 2)
newDetections.append([fixedClassNumber, detection[1], detection[2]])
if detection[0] in self.torpedoDictionary or detection[0] == "Corner":
if detection[0] != "Corner":
pass
self.imgPoints.append((detection[2][0], detection[2][1]))
self.srcPoints.append(self.torpedoSrcPoints[self.torpedoDictionary[detection[0]]])
elif detection[0] == "Corner":
pixelError = 20
'''cv2.circle(img, (int(detection[2][0]), int(detection[2][1])), 10, (255,0,0), -1)
while True:
cv2.imshow("Vision", img)
if cv2.waitKey(1) == ord(" "):
break'''
for secondDet in detections:
'''cv2.rectangle(img, (int(secondDet[2][0] - (.5 * secondDet[2][2]) - pixelError ),int(secondDet[2][1] - (.5 * secondDet[2][3]) - pixelError)), (int(secondDet[2][0] - (.5 * secondDet[2][2]) + pixelError),int(secondDet[2][1] - (.5 * secondDet[2][3]) + pixelError)), (255, 0, 0), 3)
cv2.rectangle(img, (int(secondDet[2][0]-(.5 * secondDet[2][2])), int(secondDet[2][1]- (.5 * secondDet[2][3]))), (int(secondDet[2][0] + (.5*secondDet[2][2])), int(secondDet[2][1] + (.5*secondDet[2][3]))), (0,0,255))
while True:
cv2.imshow("Vision", img)
if cv2.waitKey(1) == ord(" "):
break'''
if secondDet[0] in self.torpedoDictionary:
index = None
if detection[2][0] >= secondDet[2][0] - (.5 * secondDet[2][2]) - pixelError and detection[2][0] <= secondDet[2][0] - (.5 * secondDet[2][2]) + pixelError:
if detection[2][1] <= secondDet[2][1] - (.5 * secondDet[2][3]) + pixelError and detection[2][1] > secondDet[2][1] - (.5 * secondDet[2][3]) - pixelError:
cv2.circle(img, (int(detection[2][0]), int(detection[2][1])), 10, (255,0,0), -1)
index = self.torpedoDictionary[secondDet[0]]
index += 9
elif detection[2][0] >= secondDet[2][0] + (.5 * secondDet[2][2]) - pixelError and detection[2][0] <= secondDet[2][0] + (.5 * secondDet[2][2]) + pixelError:
if detection[2][1] <= secondDet[2][1] - (.5 * secondDet[2][3]) + pixelError and detection[2][1] > secondDet[2][1] - (.5 * secondDet[2][3]) - pixelError:
cv2.circle(img, (int(detection[2][0]), int(detection[2][1])), 10, (0,255,0), -1)
index = self.torpedoDictionary[secondDet[0]]
index += 6
if detection[2][0] >= secondDet[2][0] - (.5 * secondDet[2][2]) - pixelError and detection[2][0] <= secondDet[2][0] - (.5 * secondDet[2][2]) + pixelError:
if detection[2][1] <= secondDet[2][1] + (.5 * secondDet[2][3]) + pixelError and detection[2][1] > secondDet[2][1] + (.5 * secondDet[2][3]) - pixelError:
cv2.circle(img, (int(detection[2][0]), int(detection[2][1])), 10, (0,0,255), -1)
index = self.torpedoDictionary[secondDet[0]]
index += 8
elif detection[2][0] >= secondDet[2][0] + (.5 * secondDet[2][2]) - pixelError and detection[2][0] <= secondDet[2][0] + (.5 * secondDet[2][2]) + pixelError:
if detection[2][1] <= secondDet[2][1] + (.5 * secondDet[2][3]) + pixelError and detection[2][1] > secondDet[2][1] + (.5 * secondDet[2][3]) - pixelError:
cv2.circle(img, (int(detection[2][0]), int(detection[2][1])), 10, (255,0,255), -1)
index = self.torpedoDictionary[secondDet[0]]
index += 7
if index == None:
pass
else:
self.srcPoints.append(self.torpedoSrcPoints[index])
self.imgPoints.append((int(detection[2][0]), int(detection[2][1])))
if len(self.imgPoints) >= 4:
#print self.imgPoints, self.srcPoints
rvec, tvec = cv2.solvePnP(np.array(self.srcPoints).astype("float32"), np.array(self.imgPoints).astype("float32"), np.array(self.camMat).astype("float32"), np.zeros((4,1)))[-2:]
(pose1, jacobian) = cv2.projectPoints(np.array([(0.0,0.0,0.1)]), rvec, tvec, np.array(self.camMat).astype("float32"), None)
(pose, jacobian) = cv2.projectPoints(np.array([(0,0,12.0)]), rvec, tvec, np.array(self.camMat).astype("float32"), None)
cv2.line(img, (int(pose1[0][0][0]), int(pose1[0][0][1])), (int(pose[0][0][0]), int(pose[0][0][1])), (255,0,0), 2)
for detection in detections:
loc = detection[2]
cv2.rectangle(img, (int(loc[0]-(.5 * loc[2])), int(loc[1]- (.5 * loc[3]))), (int(loc[0] + (.5*loc[2])), int(loc[1] + (.5*loc[3]))), (0,0,255))
self.getList.append(newDetections)
cv2.imshow("Vision", img)
self.activeCamera.stopCapture()
dn.set_gpu(0)
net = dn.load_net("6_15_test.cfg", "6_15_test_40000.weights", 0)
meta = dn.load_meta("test.data")
bus = pc2.BusManager()
cam = pc2.Camera()
cam.connect(bus.getCameraFromIndex(0))
cam.startCapture()
startTime = time.time()
while True:
print "Time was", time.time() - startTime
startTime = time.time()
image = cam.retrieveBuffer()
image = image.convert(pc2.PIXEL_FORMAT.BGR)
img = np.array(image.getData(), dtype="uint8").reshape(
(image.getRows(), image.getCols(), 3))
yoloImage = dn.IMAGE()
detections = dn.detect_np(net, meta, img)
for detection in detections:
loc = detection[2]
cv2.rectangle(
img, (int(loc[0] - (.5 * loc[2])), int(loc[1] - (.5 * loc[3]))),
(int(loc[0] + (.5 * loc[2])), int(loc[1] + (.5 * loc[3]))),
(0, 0, 255))
cv2.imshow("Test", img)
key = cv2.waitKey(1)
if key == ord("q"):
break