Why is Brox Optical Flow so blotchy?
I'm getting strange results using the GPU Brox Optical Flow calculation through Emgu. Has anybody done this recently?
When the optical flow values are consistent across the image, the resulting flow values are approximately correct. When the optical flow changes in segments across the image (moving objects), the resulting optical flow values look incorrect. The optical flow image is inconsistent and scattered in areas where it should have a single value due to a moving object.
Here are the results from OpenCV Brox optical flow on GPU:
And here are the results from my own CPU-based implementation:
I've tried varying different parameters in the Brox calculation but nothing seems to help. Any idea what could be going wrong?
Emgu is NOT OpenCV but a wrapper. No way in telling if they are using the latest source, so no way to compare both setups. IF you want Emgu to get fixed, go to the proper forums.
Thanks Steve. I understand that Emgu is a wrapper. In fact there is a way of telling if they are using the latest source: look at their repo. They are using version 3.0.0. This is also stated on their main page. I don't "want Emgu to get fixed", I'm trying to understand what's going on with the Brox optical flow calculation. It's not clear whether this is an Emgu or OpenCV issue which is the main reason I'm posting here. Someone who is actually familiar with the optical flow modules would likely recognize whether this is an issue related to the optical-flow calculation parameters or something else.
Is their an option to generate Emgu wrappers yourself on latest master code? I do know from experience that 3.0.0 is already quite behind the current master in relation to fixes :)
Maybe I'll give that a shot... I don't think there is an automatic way of generating the wrapper, it would probably come down to manual conversion.