Duo Wang
Duo Wang
@smartkiwi @kashif @ezyang @djsutherland Can you help me?
@datumbox When I was training, I called [DB/backbones/resnet.py L298](https://github.com/MhLiao/DB/blob/9505b9d8f884a58fe90b884ce239a37f3b1cf404/backbones/resnet.py?_pjax=%23js-repo-pjax-container%2C%20div%5Bitemtype%3D%22http%3A%2F%2Fschema.org%2FSoftwareSourceCode%22%5D%20main%2C%20%5Bdata-pjax-container%5D#L298)，and then replaced L55 and L126 with `from torchvision.ops import DeformConv2d as ModulatedDeformConv`。 Please help me take a look. If you need...
Because the original code in the DB uses the cuda version of dcn and I only have a CPU, I want to use the dcn operator in torchvision.
The same question, I also meet same problem.
@Ixiaohuihuihui Hi, I test imgs use your codes. However the render results is bad. Do you know the reason of this cases? Thanks.
How to convert it into online processing?
> How to convert it into online processing? Can you give some guidance?
@Puzer I have the same question. Looking forward to your reply!
The solution is to convert multi-class labels to binary labels.