https://github.com/charlesshang/dcnv2

Deformable Convolutional Networks v2 with Pytorch
https://github.com/charlesshang/dcnv2

cnn deformable-convolutional-networks pytorch

Last synced: 3 days ago
JSON representation

Deformable Convolutional Networks v2 with Pytorch

• Host: GitHub
• URL: https://github.com/charlesshang/dcnv2
• Owner: CharlesShang
• License: bsd-3-clause
• Created: 2018-12-05T05:16:57.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2024-04-25T06:55:55.000Z (8 months ago)
• Last Synced: 2024-12-20T10:05:12.734Z (3 days ago)
• Topics: cnn, deformable-convolutional-networks, pytorch
• Language: C++
• Size: 1.38 MB
• Stars: 1,293
• Watchers: 24
• Forks: 400
• Open Issues: 111
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
## Deformable Convolutional Networks V2 with Pytorch 1.X

### Build

```bash

    ./make.sh         # build

    python testcpu.py    # run examples and gradient check on cpu

    python testcuda.py   # run examples and gradient check on gpu 

```

### Note

Now the master branch is for pytorch 1.x, you can switch back to pytorch 0.4 with,

```bash

git checkout pytorch_0.4

```

### Known Issues:

- [x] Gradient check w.r.t offset (solved)

- [ ] Backward is not reentrant (minor)

This is an adaption of the official [Deformable-ConvNets](https://github.com/msracver/Deformable-ConvNets/tree/master/DCNv2_op).

Update: all gradient check passes with **double** precision. 

Another issue is that it raises `RuntimeError: Backward is not reentrant`. However, the error is very small (`<1e-7` for 

float `<1e-15` for double), 

so it may not be a serious problem (?)

Please post an issue or PR if you have any comments.