CBNet: A Novel Composite Backbone Network Architecture for Object Detection

by Yudong Liu

Update

We release our code based on mmdetection, and you can refer to https://github.com/VDIGPKU/CBNet

Introduction

This repository is modified on Cascade R-CNN, which is re-implemented by by Zhaowei Cai and Nuno Vasconcelos on the base of Detectron.

Please follow Detectron on how to install the environment. The project is only free for academic research purposes. For commerce permission, please contact wyt@pku.edu.cn.

Citation

If you use our code/model/data, please cite our paper: https://aaai.org/Papers/AAAI/2020GB/AAAI-LiuY.1833.pdf

Cascade Mask R-CNN without Bells & Whistles

backbone	type	lr schd	im/ gpu	box AP	box AP50	box AP75	mask AP	mask AP50	mask AP75
[X-152-32x8d-FPN-IN5k-cascade]	Mask	s1x	1	48.3	67.0	52.8	41.0	64.1	44.2
Dual-X-152-32x8d-FPN-IN5k-cascade	Mask	s1x	1	50.0	68.8	54.6	42.0	64.6	45.6