Haofei Xu
Haofei Xu
Hi all, sorry for the late response. If this issue is still relavant to you, I would suggest to try our new GMStereo model: https://haofeixu.github.io/unimatch/ & https://github.com/autonomousvision/unimatch. No CUDA op...
Sorry that the training log is no longer available due to the long time of this project. I would suggest you to first try our model on our dataset, and...
Hi @Estrella1024 , you can save the prediction results on KITTI test set and submit the results to the online server.
Hi all, sorry for the late response. If this issue is still relavant to you, I would suggest to try our new GMStereo model: https://haofeixu.github.io/unimatch/ & https://github.com/autonomousvision/unimatch. No CUDA op...
Hi @schc54 , can you please give more details? The output ought to be non-negative.
Which dataset are you using? In the datasets we used in our paper, we didn't observe negative output. Could you attach the training log here?
Hi @tholmb , I have released the details and code regarding how to extend GMFlow to stereo, you can find the info at https://haofeixu.github.io/unimatch/ and https://github.com/autonomousvision/unimatch, thanks.
Thanks for your interest in our work. To save the .flo flow results, you can use the `save_flow_flo` argument: https://github.com/haofeixu/gmflow/blob/main/evaluate.py#L572
1. No, you should use the raw .flo results for warping 2. Direct load the .flo flow file to get the uv vector 3. Please refer to this function https://github.com/haofeixu/gmflow/blob/main/gmflow/geometry.py#L65...