Mask Encoding for Single Shot Instance Segmentation

Rufeng Zhang, Zhi Tian, Chunhua Shen, Mingyu You, Youliang Yan

[arXiv] [BibTeX]

Models

COCO Instance Segmentation Baselines with MEInst

Name	inf. time	box AP	mask AP	download
MEInst_R_50_1x_none	13 FPS	39.5	30.7	model
MEInst_R_50_1x	12 FPS	40.1	31.7	model
MEInst_R_50_3x	12 FPS	43.6	34.5	model
MEInst_R_50_3x_512	19 FPS	40.8	32.2	model

Inference time is measured on a NVIDIA 1080Ti with batch size 1.

Quick Start

1. Download the matrix file for mask encoding during training
2. Symlink the matrix path to datasets/components/xxx.npz, e.g., coco/components/coco_2017_train_class_agnosticTrue_whitenTrue_sigmoidTrue_60.npz
3. Follow AdelaiDet for install, train and inference

Step by step for Mask Encoding (Optional)

We recommend to directly download the matrix file and use it, as it can already handle most cases. And we also provide tools to generate encoding matrix yourself.

Example:

• Generate encoding matrix

  python adet/modeling/MEInst/LME/mask_generation.py

• Evaluate the quality of reconstruction

  python adet/modeling/MEInst/LME/mask_evaluation.py

Citing MEInst

If you use MEInst, please use the following BibTeX entry.

@inproceedings{zhang2020MEInst,
  title     =  {Mask Encoding for Single Shot Instance Segmentation},
  author    =  {Zhang, Rufeng and Tian, Zhi and Shen, Chunhua and You, Mingyu and Yan, Youliang},
  booktitle =  {Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR)},
  year      =  {2020}
}

License

For academic use, this project is licensed under the 2-clause BSD License - see the LICENSE file for details. For commercial use, please contact Chunhua Shen.