Towards Adaptive Open-Set Object Detection via Category-Level Collaboration Knowledge Mining

2026-04-13

AI Summary

This work addresses the limitations of existing adaptive open-set object detection methods, which suffer from weak cross-domain representations, ambiguity in novel categories, and source-domain feature bias under the absence of target-domain annotations. To overcome these challenges, the authors propose a category-level collaborative knowledge mining framework that constructs a memory bank via clustering to encode both category prototypes and intra-class variations. A selection metric is designed to initialize novel-class classifiers by transferring knowledge from base classes, coupled with an adaptive feature assignment mechanism for effective knowledge transfer. Innovatively, the method integrates inter- and intra-class relationships across domains and mitigates source-domain bias through an iteratively updated memory bank and an asynchronous optimization strategy. Experiments demonstrate that the proposed approach outperforms state-of-the-art methods by 1.1–5.5 mAP on multiple benchmarks, significantly enhancing cross-domain novel category discovery and detection performance.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Object Detection & Categorization
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Vertical and domain-specific search

Abstract

Existing object detectors often struggle to generalize across domains while adapting to emerging novel categories. Adaptive open-set object detection (AOOD) addresses this challenge by training on base categories in the source domain and adapting to both base and novel categories in the target domain without target annotations. However, current AOOD methods remain limited by weak cross-domain representations, ambiguity among novel categories, and source-domain feature bias. To address these issues, we propose a category-level collaboration knowledge mining strategy that exploits both inter-class and intra-class relationships across domains. Specifically, we construct a clustering-based memory bank to encode class prototypes, auxiliary features, and intra-class disparity information, and iteratively update it via unsupervised clustering to enhance category-level knowledge representation. We further design a base-to-novel selection metric to discover source-domain features related to novel categories and use them to initialize novel-category classifiers. In addition, an adaptive feature assignment strategy transfers the learned category-level knowledge to the target domain and asynchronously updates the memory bank to alleviate source-domain bias. Extensive experiments on multiple benchmarks show that our method consistently surpasses state-of-the-art AOOD methods by 1.1-5.5 mAP.

Problem

• Adaptive Open-Set Object Detection
• cross-domain generalization
• novel category adaptation
• source-domain bias
• unlabeled target domain

Innovation

• adaptive open-set object detection
• category-level collaboration
• memory bank
• unsupervised clustering
• feature bias mitigation