SemPLeS: Semantic Prompt Learning for Weakly-Supervised Semantic Segmentation

2024-01-22

arXiv.org

AI Summary

Weakly supervised semantic segmentation (WSSS) suffers from localization bias, as pseudo-masks generated from image-level labels often concentrate on discriminative object regions or co-occurring background, failing to capture full object extents. To address this, we propose a contrastive prompt learning framework operating in CLIP’s latent space: (1) learnable text prompts enhance category-region semantic alignment; (2) a prompt-guided semantic refinement module explicitly models and suppresses interference from category-co-occurring backgrounds. Our approach breaks the reliance on discriminative regions inherent in CAM-based methods and is the first to synergistically integrate contrastive learning, prompt tuning, and multi-stage semantic refinement. Evaluated on PASCAL VOC 2012 and MS COCO 2014, it achieves state-of-the-art performance, with substantial gains in pseudo-mask mIoU. Moreover, the framework is plug-and-play—compatible with and improving diverse WSSS pipelines without architectural modification.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Semi-Supervised Learning
• Natural Language Processing: Prompt Engineering / Prompting

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Weakly-Supervised Semantic Segmentation (WSSS) aims to train segmentation models using image data with only image-level supervision. Since precise pixel-level annotations are not accessible, existing methods typically focus on producing pseudo masks for training segmentation models by refining CAM-like heatmaps. However, the produced heatmaps may capture only the discriminative image regions of object categories or the associated co-occurring backgrounds. To address the issues, we propose a Semantic Prompt Learning for WSSS (SemPLeS) framework, which learns to effectively prompt the CLIP latent space to enhance the semantic alignment between the segmented regions and the target object categories. More specifically, we propose Contrastive Prompt Learning and Prompt-guided Semantic Refinement to learn the prompts that adequately describe and suppress the co-occurring backgrounds associated with each object category. In this way, SemPLeS can perform better semantic alignment between object regions and class labels, resulting in desired pseudo masks for training segmentation models. The proposed SemPLeS framework achieves competitive performance on standard WSSS benchmarks, PASCAL VOC 2012 and MS COCO 2014, and shows compatibility with other WSSS methods. Code: https://github.com/NVlabs/SemPLeS.

Problem

• Weakly Supervised Semantic Segmentation
• Pixel-level Annotation
• Object Recognition Accuracy

Innovation

• SemPLeS Framework
• Weakly Supervised Semantic Segmentation
• Object-Background Relationship Learning