When I use the network to train the 360monodepth database, the result is very poor. In the code, I only replace the data, and read the depth map.
The original code is as following:
depth_fd = h5py.File(depth_path, "r")
distance_meters = np.array(depth_fd['dataset'])
depth = hypersim_distance_to_depth(distance_meters)
I modity the code as following:
depth = (cv2.imread(depth_path, cv2.CV_16UC1)) / 255.0
When I use the network to train the 360monodepth database, the result is very poor. In the code, I only replace the data, and read the depth map. The original code is as following: depth_fd = h5py.File(depth_path, "r") distance_meters = np.array(depth_fd['dataset']) depth = hypersim_distance_to_depth(distance_meters) I modity the code as following: depth = (cv2.imread(depth_path, cv2.CV_16UC1)) / 255.0
