/D2Zero

[CVPR-2023] Semantic-Promoted Debiasing and Background Disambiguation for Zero-Shot Instance Segmentation

Primary LanguagePythonMIT LicenseMIT

Semantic-Promoted Debiasing and Background Disambiguation for Zero-Shot Instance Segmentation

PyTorch Python

🏠[Project page]   📄[arXiv]   📄[PDF]

This repository contains code for CVPR2023 paper:

Semantic-Promoted Debiasing and Background Disambiguation for Zero-Shot Instance Segmentation Shuting He, Henghui Ding, Wei Jiang

framework

Installation:

The code is tested under CUDA 11.2, Pytorch 1.9.0 and Detectron2 0.6

  1. Follow the installation process of Mask2Former
  2. Install other required packages: pip -r requirements.txt
  3. Prepare the dataset following datasets/README.md

Inference

GZSIS Settings:

python train_net.py  --config-file configs/d2zero_48_17.yaml --num-gpus 8 --eval-only MODEL.WEIGHTS d2zero_48_17/model_final.pth

ZSIS Settings:

python train_net.py  --config-file configs/d2zero_48_17.yaml --num-gpus 8 --eval-only MODEL.WEIGHTS d2zero_48_17/model_final.pth DATASETS.TEST '("coco_zsi_48_17_val_unseen_only",)' MODEL.MASK_FORMER.TEST.GENERALIZED False

Training

python train_net.py  --config-file configs/d2zero_48_17.yaml --num-gpus 8

Acknowledgement

This project is based on ZSI, Mask2Former. Many thanks to the authors for their great works!

BibTeX

Please consider citing D2Zero if it helps your research.

@inproceedings{D2Zero,
  title={Semantic-Promoted Debiasing and Background Disambiguation for Zero-Shot Instance Segmentation},
  author={He, Shuting and Ding, Henghui and Jiang, Wei},
  booktitle={CVPR},
  year={2023}
}