Open VojiWest opened 6 months ago
Hi, thanks for ur attention to our work, and sorry for the late response. The implementation code for NYUv2 is still being cleaned up. If you plan to compress MTINet_hrnet18 on NYUv2, it is recommended that the channels of offset convs in DCNv2 be fully preserved during pruning since they are somehow complicated to determine the structural sparsity. Other implementations are similar to the compression of YOLOv5.
Hello, I see from your paper that you used PAGCP on the NYUv2 dataset, but in the code here you only have implementation for VOC and COCO, is it possible to get the implementation for NYUv2? If not, any tips would be appreciated. Thanks!