Towards Robust Pseudo-Label Learning in Semantic Segmentation: An Encoding Perspective

2025-12-07

AI Summary

Pseudo-labeling in unsupervised domain adaptation (UDA) and semi-supervised learning (SSL) for semantic segmentation is highly susceptible to erroneous pseudo-labels, especially due to error amplification inherent in one-hot encoding. To address this, we propose an encoding-robustness enhancement paradigm. First, we adopt Error-Correcting Output Codes (ECOC) to decouple class semantics, transforming pixel-wise classification into multi-bit-level prediction. Second, we design a bit-level pseudo-label denoising mechanism that dynamically identifies and rectifies unreliable bits during training, thereby suppressing error propagation. Our approach is architecture-agnostic and can be seamlessly integrated as a plug-and-play module into mainstream segmentation networks. Extensive experiments on multiple UDA benchmarks (e.g., GTA→Cityscapes) and SSL benchmarks demonstrate consistent and significant performance gains, validating both its generalizability and effectiveness.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Semi-Supervised Learning
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Pseudo-label learning is widely used in semantic segmentation, particularly in label-scarce scenarios such as unsupervised domain adaptation (UDA) and semisupervised learning (SSL). Despite its success, this paradigm can generate erroneous pseudo-labels, which are further amplified during training due to utilization of one-hot encoding. To address this issue, we propose ECOCSeg, a novel perspective for segmentation models that utilizes error-correcting output codes (ECOC) to create a fine-grained encoding for each class. ECOCSeg offers several advantages. First, an ECOC-based classifier is introduced, enabling model to disentangle classes into attributes and handle partial inaccurate bits, improving stability and generalization in pseudo-label learning. Second, a bit-level label denoising mechanism is developed to generate higher-quality pseudo-labels, providing adequate and robust supervision for unlabeled images. ECOCSeg can be easily integrated with existing methods and consistently demonstrates significant improvements on multiple UDA and SSL benchmarks across different segmentation architectures. Code is available at https://github.com/Woof6/ECOCSeg.

Problem

• Addresses erroneous pseudo-labels in semantic segmentation
• Improves pseudo-label quality via error-correcting output codes
• Enhances robustness in unsupervised and semi-supervised learning

Innovation

• ECOC-based classifier for class attribute disentanglement
• Bit-level label denoising for robust pseudo-labels
• Integration with existing segmentation methods for UDA/SSL