SG-Net: Spatial Granularity Network for One-Stage Video Instance Segmentation

Introduction

We approach the [VIS task] from a new perspective and propose a one-stage spatial granularity network (SG-Net) (as shown in the above figure). This repo is the implementation of CVPR 2021 paper "SG-Net: Spatial Granularity Network for One-Stage Video Instance Segmentation." [pdf]

Installation

Please find detailed steps Here for installation and dataset preparation.

Train

Please find details Here for step-by-step instructions.

Inference

Please refer to Here for inference.

License

SG-Net is released under the MIT license.

Citation

If you find this repo useful for your research, please consider citing the paper

@article{liu2021sg,
  title={SG-Net: Spatial Granularity Network for One-Stage Video Instance Segmentation},
  author={Liu, Dongfang and Cui, Yiming and Tan, Wenbo and Chen, Yingjie},
  journal={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2021}
}

Acknowledgements

We truely thanksful of the following piror efforts in terms of knowledge contributions and open-source repos.

BlendMask: Top-down meets bottom-up for instance segmentation (CVPR'20) [paper] [official code]
Video Instance Segmenation (ICCV'19) [paper] [official code]

goodproj13 / SG-Net

readme