Dual-Thresholding Heatmaps to Cluster Proposals for Weakly Supervised Object Detection

2025-09-10

AI Summary

To address three key challenges in weakly supervised object detection (WSOD)—incomplete pseudo-labels, intra-class instance confusion, and slow convergence—this paper proposes a heatmap-guided dual-threshold clustering framework. First, a dual-threshold heatmap mechanism is designed alongside candidate box clustering to jointly improve localization completeness and instance separability. Second, explicit background class representation and heatmap-based pre-supervision are integrated into the WSBDN architecture to reduce semantic discrepancy between branches. Third, a negative certainty supervision loss is introduced to effectively leverage ignored proposals and accelerate convergence. Evaluated on PASCAL VOC 2007 and 2012, the method achieves mean Average Precision (mAP) of 58.5% and 55.6%, and mean Correct Localization (mCorLoc) of 81.8% and 80.5%, respectively—significantly outperforming state-of-the-art approaches.

Technology Category

• Computer Vision: Object Detection & Categorization
• Machine Learning: Semi-Supervised Learning
• Search and Optimization: Learning to Search

Application Category

• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Weakly supervised object detection (WSOD) has attracted significant attention in recent years, as it does not require box-level annotations. State-of-the-art methods generally adopt a multi-module network, which employs WSDDN as the multiple instance detection network module and multiple instance refinement modules to refine performance. However, these approaches suffer from three key limitations. First, existing methods tend to generate pseudo GT boxes that either focus only on discriminative parts, failing to capture the whole object, or cover the entire object but fail to distinguish between adjacent intra-class instances. Second, the foundational WSDDN architecture lacks a crucial background class representation for each proposal and exhibits a large semantic gap between its branches. Third, prior methods discard ignored proposals during optimization, leading to slow convergence. To address these challenges, we first design a heatmap-guided proposal selector (HGPS) algorithm, which utilizes dual thresholds on heatmaps to pre-select proposals, enabling pseudo GT boxes to both capture the full object extent and distinguish between adjacent intra-class instances. We then present a weakly supervised basic detection network (WSBDN), which augments each proposal with a background class representation and uses heatmaps for pre-supervision to bridge the semantic gap between matrices. At last, we introduce a negative certainty supervision loss on ignored proposals to accelerate convergence. Extensive experiments on the challenging PASCAL VOC 2007 and 2012 datasets demonstrate the effectiveness of our framework. We achieve mAP/mCorLoc scores of 58.5%/81.8% on VOC 2007 and 55.6%/80.5% on VOC 2012, performing favorably against the state-of-the-art WSOD methods. Our code is publicly available at https://github.com/gyl2565309278/DTH-CP.

Problem

• Addresses pseudo GT boxes focusing on parts not full objects
• Resolves lack of background class representation in WSDDN
• Solves slow convergence from discarding ignored proposals

Innovation

• Heatmap-guided proposal selector with dual thresholds
• Weakly supervised detection network with background class
• Negative certainty supervision loss for ignored proposals