Skip to content

idstcv/DR_loss

master
Switch branches/tags
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 

DR Loss

PyTorch Implementation for Our CVPR'20 Paper: "DR Loss: Improving Object Detection by Distributional Ranking"

Requirements

Usage:

  1. Put the loss file to the codebase of maskrcnn_benchmark at
maskrcnn-benchmark/maskrcnn_benchmark/layers/sigmoid_dr_loss.py

and add the class into "init.py".

  1. Change the focal loss in RetinaNet to the dr loss at
maskrcnn-benchmark/maskrcnn_benchmark/modeling/rpn/retinanet/loss.py
  1. Run RetinaNet with the configurations in "configs/dr_retina".

Models:

model lr sched multi-scale training mAP(minival) mAP (test-dev) link
Dr_Retina_R-50-FPN 1x No 37.4 37.6 Google Drive
Dr_Retina_R-101-FPN 2x Yes 41.5 41.7 Google Drive

Citation

If you use the package in your research, please cite our paper:

@inproceedings{qian2020dr,
  author    = {Qi Qian and
               Lei Chen and
               Hao Li and
               Rong Jin},
  title     = {DR Loss: Improving Object Detection by Distributional Ranking},
  booktitle = {{IEEE} Conference on Computer Vision and Pattern Recognition, {CVPR} 2020},
  year      = {2020}
}

About

PyTorch Implementation for DR Loss

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages