By Yi Li, Zhanghui Kuang, Liyang Liu, Yimin Chen, Wayne Zhang
SenseTime, Tsinghua University
This is a PyTorch implementation of Pseudo-mask Matters in Weakly-supervised Semantic Segmentation.(ICCV2021).
In this paper, we propose Coefficient of Variation Smoothing and Proportional Pseudo-mask Generation to generate high quality pseudo-mask in classification part. In segmentation part, we propose Pretended Under-Fitting strategy and Cyclic Pseudo-mask for better utilization of pseudo-mask.
- Download VOC12 OneDrive, BaiduYun
- Download COCO14 BaiduYun
- Download pretrained models OneDrive, BaiduYun
(extract code of BaiduYun: mtci)
git clone https://github.com/Eli-YiLi/PMM cd PMM ln -s [path to model files] models ln -s [path to VOC12] voc12 ln -s [path to COCO14] coco14 pip3 install -r requirements.txt bash slurm_run.sh [partition name] [dataset name] / bash dist_run.sh [dataset name]
Please refer to WSSS_MMSeg
Please refer to: LICENSE.