ST++: Make Self-trainingWork Better for Semi-supervised Semantic Segmentation
Authors:
Top Cited Papers
- 1 June 2022
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 4258-4267
Abstract
Self-training via pseudo labeling is a conventional, simple, and popular pipeline to leverage unlabeled data. In this work, we first construct a strong baseline of self-training (namely ST) for semi-supervised semantic segmentation via injecting strong data augmentations (SDA) on unlabeled images to alleviate overfitting noisy labels as well as decouple similar predictions between the teacher and student. With this simple mechanism, our ST outperforms all existing methods without any bells and whistles, e.g., iterative retraining. Inspired by the impressive results, we thoroughly investigate the SDA and provide some empirical analysis. Nevertheless, incorrect pseudo labels are still prone to accumulate and degrade the performance. To this end, we further propose an advanced self-training framework (namely ST++), that performs selective re-training via prioritizing reliable unlabeled images based on holistic prediction-level stability. Concretely, several model checkpoints are saved in the first stage supervised training, and the discrepancy of their predictions on the unlabeled image serves as a measurement for reliability. Our image-level selection offers holistic contextual information for learning. We demonstrate that it is more suitable for segmentation than common pixel-wise selection. As a result, ST+ further boosts the performance of our ST. Code is available at https://github.com/LiheYoung/ST-PlusPlus.Keywords
All Related Versions
Funding Information
- National Key Research and Development Program of China (2019YFC0118300)
- NSFC major program (62192783)
- China Postdoctoral Science Foundation (2021M690609)
- Jiangsu Natural Science Foundation (BK20210224)
This publication has 25 references indexed in Scilit:
- Semi-supervised Semantic Segmentation with Directional Context-aware Consistency2021
- Semi-Supervised Semantic Segmentation With Cross-Consistency Training2020
- Not All Areas Are Equal: Transfer Learning for Semantic Segmentation via Hierarchical Region Selection2019
- Data Distillation: Towards Omni-Supervised Learning2018
- Not All Pixels Are Equal: Difficulty-Aware Semantic Segmentation via Deep Layer Cascade2017
- Pyramid Scene Parsing Network2017
- Deep Residual Learning for Image Recognition2016
- The Cityscapes Dataset for Semantic Urban Scene Understanding2016
- The Pascal Visual Object Classes Challenge: A RetrospectiveInternational Journal of Computer Vision, 2014
- ImageNet: A large-scale hierarchical image database2009