cam.set(cv2.CAP_PROP_FRAME_HEIGHT, frame_height)
while(True):
## Used for writing file to matlab
#file = open('classFile.txt','a')
# Capture frame by frame
_, frame = cam.read()
# Perform radial undistortion (Pupil world camera has significant radial distortion which adversely affects detection/classification).
frame = cv2.undistort(frame, camera_matrix, dist_coefs)
cv2.imshow('Object Detection',frame)
# Uses neural network (darkflow) to predict detected objects and respective classifications.
objects_detected = tfnet.return_predict2(frame)
#print(objects_detected)
# Collect normalized gaze_data, denormalize according to camera resolution (bottom-left corner is (0,0), top-right is (1,1)).
gaze_data = getGazeData()
gaze_x = (gaze_data['gaze_coord'][0])*frame_width
gaze_y = (1-gaze_data['gaze_coord'][1])*frame_height # Y-coordinate was flipped initially
# if gaze_data['confidence']>.7:
# print(gazeX, gazeY, ' confidence:', gaze_data['confidence'])
# Append gaze point to real-time stream as a green dot.
frame = cv2.circle(frame, (int(gaze_x), int(gaze_y)), 10, (0,255,0), -1)
cv2.imshow('Object Detection',frame)
if objects_detected:
