tomrunia / opticalflow_visualization Goto Github PK

It would be great if you could mention how colorwheel can be visualized . As i see this is the array of colors but confused if its rgb or hue and how can we have the image generated using it.

there is no .npy file in data folder

Hi,
Thanks for this amazing tool! I have a question about flow color. According to the flow color coding, it seems that we should set 'convert_to_bgr=True' to match flow color coding? Am I right?

This is the issue When I run this code:
Traceback (most recent call last):
File "ceshi.py", line 44, in
flow_color = flow_vis.flow_to_color(flow_uv, convert_to_bgr=False)
File "/home/fang/OpticalFlow_Visualization-master/flow_vis.py", line 142, in flow_to_color
return flow_compute_color(u, v, convert_to_bgr)
File "/home/fang/OpticalFlow_Visualization-master/flow_vis.py", line 99, in flow_compute_color
col0 = tmp[k0] / 255.0
IndexError: index 55 is out of bounds for axis 0 with size 55

Can somebody tell me What's wrong about it?

Hi, great repo!
It would be helpful to add a few words about the format of the input to the flow_to_color.
The dostring ("flow_uv (np.ndarray): Flow UV image of shape [H,W,2]") only mentions the required shape, but what about the values?
I guess it's the relative shift (and not absolute position)? are the units in "pixels" or should I divide by H,W?
also, if clip_flow parameter is passed, the input flow is clipped between [0, clip_flow], which means negative values are not expected? which means the input should not be the relative shift, but then the output makes no sense.

It could have also helped,If the example numpy data ('./data/flow_example_data.npy') was available