Semi-supervised semantic segmentation needs strong, varied perturbations

Abstract

Consistency regularization describes a class of approaches that have yielded ground breaking results in semi-supervised classification problems. Prior work has established the cluster assumption - under which the data distribution consists of uniform class clusters of samples separated by low density regions - as important to its success. We analyze the problem of semantic segmentation and find that its' distribution does not exhibit low density regions separating classes and offer this as an explanation for why semi-supervised segmentation is a challenging problem, with only a few reports of success. We then identify choice of augmentation as key to obtaining reliable performance without such low-density regions. We find that adapted variants of the recently proposed CutOut and CutMix augmentation techniques yield state-of-the-art semi-supervised semantic segmentation results in standard datasets. Furthermore, given its challenging nature we propose that semantic segmentation acts as an effective acid test for evaluating semi-supervised regularizers. Implementation at: https://github.com/Britefury/cutmix-semisup-seg.

arXiv comment21 pages, 7 figures, accepted to BMVC 2020

References in corpus (10)

• Rethinking Atrous Convolution for Semantic Image Segmentation
• Improved Regularization of Convolutional Neural Networks with Cutout
• FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence
• Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results
• Interpolation Consistency Training for Semi-Supervised Learning
• MixMatch: A Holistic Approach to Semi-Supervised Learning
• A DIRT-T Approach to Unsupervised Domain Adaptation
• Adversarial Learning for Semi-Supervised Semantic Segmentation
• ReMixMatch: Semi-Supervised Learning with Distribution Alignment and Augmentation Anchoring
• There Are Many Consistent Explanations of Unlabeled Data: Why You Should Average

Cited by in corpus (13)

• Self-Training: A Survey
• Learning Self-Supervised Low-Rank Network for Single-Stage Weakly and Semi-Supervised Semantic Segmentation
• Uncertainty-Aware Consistency Regularization for Cross-Domain Semantic Segmentation
• Consistency-Regularized Region-Growing Network for Semantic Segmentation of Urban Scenes with Point-Level Annotations
• Conservative-Progressive Collaborative Learning for Semi-supervised Semantic Segmentation
• PIT: Position-Invariant Transform for Cross-FoV Domain Adaptation
• Domain Adaptive Semantic Segmentation via Regional Contrastive Consistency Regularization
• Semi-Supervised Building Footprint Generation with Feature and Output Consistency Training
• Semi-Supervised Semantic Segmentation via Adaptive Equalization Learning
• Condition-Invariant Semantic Segmentation
• Revisiting consistency for semi-supervised semantic segmentation
• Boundary-Refined Prototype Generation: A General End-to-End Paradigm for Semi-Supervised Semantic Segmentation
• Cross-Domain Transfer Learning with CoRTe: Consistent and Reliable Transfer from Black-Box to Lightweight Segmentation Model