Robust Mutual Learning for Semi-supervised Semantic Segmentation
arXiv:2106.00609
Abstract
Recent semi-supervised learning (SSL) methods are commonly based on pseudo labeling. Since the SSL performance is greatly influenced by the quality of pseudo labels, mutual learning has been proposed to effectively suppress the noises in the pseudo supervision. In this work, we propose robust mutual learning that improves the prior approach in two aspects. First, the vanilla mutual learners suffer from the coupling issue that models may converge to learn homogeneous knowledge. We resolve this issue by introducing mean teachers to generate mutual supervisions so that there is no direct interaction between the two students. We also show that strong data augmentations, model noises and heterogeneous network architectures are essential to alleviate the model coupling. Second, we notice that mutual learning fails to leverage the network's own ability for pseudo label refinement. Therefore, we introduce self-rectification that leverages the internal knowledge and explicitly rectifies the pseudo labels before the mutual teaching. Such self-rectification and mutual teaching collaboratively improve the pseudo label accuracy throughout the learning. The proposed robust mutual learning demonstrates state-of-the-art performance on semantic segmentation in low-data regime.
References in corpus (14)
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Improved Regularization of Convolutional Neural Networks with Cutout
- Temporal Ensembling for Semi-Supervised Learning
- Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification
- Billion-scale semi-supervised learning for image classification
- PseudoSeg: Designing Pseudo Labels for Semantic Segmentation
- ReMixMatch: Semi-Supervised Learning with Distribution Alignment and Augmentation Anchoring
- Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation
- SelfMatch: Combining Contrastive Self-Supervision and Consistency for Semi-Supervised Learning
- Naive-Student: Leveraging Semi-Supervised Learning in Video Sequences for Urban Scene Segmentation
- A Three-Stage Self-Training Framework for Semi-Supervised Semantic Segmentation
- A Simple Baseline for Semi-supervised Semantic Segmentation with Strong Data Augmentation
- Mask-based Data Augmentation for Semi-supervised Semantic Segmentation