ST++: Make Self-training Work Better for Semi-supervised Semantic Segmentation

TL;DR

This paper enhances semi-supervised semantic segmentation by improving self-training with strong data augmentations and a novel reliability-based re-training strategy, significantly outperforming existing methods.

Contribution

Introduces a strong baseline for self-training with data augmentation and proposes ST++ with holistic prediction stability for better pseudo-label reliability.

Findings

ST outperforms existing methods without iterative re-training.

SDA effectively reduces overfitting and prediction similarity.

ST++ further improves performance through reliability-based selective re-training.

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 re-training. 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.