Skip to content

SHI-Labs/Pseudo-IoU-for-Anchor-Free-Object-Detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Pseudo-IoU-for-Anchor-Free-Object-Detection

This is the repo to host the code for Pseudo-IoU in the following paper: Arxiv link

By Jiachen Li, Bowen Cheng, Rogerio Feris, Jinjun Xiong, Thomas S.Huang, Wen-Mei Hwu and Humphrey Shi.

Introduction

Current anchor-free object detectors are quite simple and effective yet lack accurate label assignment methods, which limits their potential in competing with classic anchor-based models that are supported by welldesigned assignment methods based on the Intersectionover-Union (IoU) metric. In this paper, we present Pseudo Intersection-over-Union (Pseudo-IoU): a simple metric that brings more standardized and accurate assignment rule into anchor-free object detection frameworks without any additional computational cost or extra parameters for training and testing, making it possible to further improve anchor-free object detection by utilizing training samples of good quality under effective assignment rules that have been previously applied in anchor-based methods. By incorporating Pseudo-IoU metric into an end-toend single-stage anchor-free object detection framework, we observe consistent improvements in their performanceon general object detection benchmarks such as PASCAL VOC and MSCOCO. Our method (single-model and singlescale) also achieves comparable performance to other recent state-of-the-art anchor-free methods without bells and whistles.

Prerequisites

  • Python 3.7
  • PyTorch 1.7.0
  • CUDA 11.0
  • MMdetection v2.11.0

Please following the installation of mmdetection and merges Pseudo-IoU configs and models into mmdetection folder.

Results

More models will be released soon

Backbone Lr schd box_mAP box_mAP_50 box_mAP_75 box_mAP_s box_mAP_m box_mAP_l Config Download
R-50 1x 38.4 57.4 40.9 23.8 42.5 48.8 config model
R-101 1x 40.4 59.5 40.9 23.7 44.9 51.4 config model
R-101-DCN 2x 43.5 62.9 46.6 25.7 47.4 57.6 config model

Citation

@article{li2021pseudoiou,
  title={Pseudo-IoU: Improving Label Assignment in Anchor-Free Object Detection},
  author={Jiachen Li, Bowen Cheng, Rogerio Feris, Jinjun Xiong, Thomas S.Huang, Wen-Mei Hwu and Humphrey Shi},
  journal={arXiv preprint arXiv:2104.14082},
  year={2021}
}

About

Pseudo-IoU: Improving Label Assignment in Anchor-Free Object Detection

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages