Ping Hu

Hi, thanks for your interest. The key idea is to separate and distribute a large segmentation model's computation into multiple independent lightweight subnetworks, which can be assigned to different frames...

Thanks for your interest. We plan to clean up and release the code after the cvpr deadline.

Hi, this code is just for speed evaluation, so the BN layers are removed as they can be merged into the Conv layers before them.

Yes, it should be. Yet that doesn't affect the efficiency.

Hi, can you try this https://github.com/feinanshan/TDNet/issues/23#issuecomment-727599081

Hi, this is due to the fixed input size of Layer_Norm. To train/test with a different input size, you can resize the weights of Layer_Norm in the pretrained model into...