MaochengHu 576cda45b8 first commit 2 years ago
..
_base_ 576cda45b8 first commit 2 years ago
README.md 576cda45b8 first commit 2 years ago
gfl_r101vd_fpn_mstrain_2x_coco.yml 576cda45b8 first commit 2 years ago
gfl_r18vd_1x_coco.yml 576cda45b8 first commit 2 years ago
gfl_r34vd_1x_coco.yml 576cda45b8 first commit 2 years ago
gfl_r50_fpn_1x_coco.yml 576cda45b8 first commit 2 years ago
gflv2_r50_fpn_1x_coco.yml 576cda45b8 first commit 2 years ago

README.md

Generalized Focal Loss Model(GFL)

Introduction

We reproduce the object detection results in the paper Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection and Generalized Focal Loss V2. And We use a better performing pre-trained model and ResNet-vd structure to improve mAP.

Model Zoo

Backbone Model batch-size/GPU lr schedule FPS Box AP download config
ResNet50 GFL 2 1x ---- 41.0 model | log config
ResNet101-vd GFL 2 2x ---- 46.8 model | log config
ResNet34-vd GFL 2 1x ---- 40.8 model | log config
ResNet18-vd GFL 2 1x ---- 36.6 model | log config
ResNet50 GFLv2 2 1x ---- 41.2 model | log config

Notes:

  • GFL is trained on COCO train2017 dataset with 8 GPUs and evaluated on val2017 results of mAP(IoU=0.5:0.95).

Citations

@article{li2020generalized,
  title={Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection},
  author={Li, Xiang and Wang, Wenhai and Wu, Lijun and Chen, Shuo and Hu, Xiaolin and Li, Jun and Tang, Jinhui and Yang, Jian},
  journal={arXiv preprint arXiv:2006.04388},
  year={2020}
}

@article{li2020gflv2,
  title={Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection},
  author={Li, Xiang and Wang, Wenhai and Hu, Xiaolin and Li, Jun and Tang, Jinhui and Yang, Jian},
  journal={arXiv preprint arXiv:2011.12885},
  year={2020}
}