Proto-OOD: Enhancing OOD Object Detection with Prototype Feature Similarity

2024-09-09

arXiv.org

AI Summary

To address the problem of out-of-distribution (OOD) object misclassification in open-world object detection, this paper proposes a prototype-based feature similarity framework for OOD detection. Methodologically, it introduces the first joint integration of prototype learning and contrastive learning for OOD discrimination; designs a negative-sample embedding generator to enhance diversity in OOD training data; and establishes a more realistic, deployment-oriented OOD evaluation protocol. Under the cross-domain setting—using Pascal VOC as the in-distribution (ID) dataset and MS-COCO as the OOD dataset—the method significantly reduces false positive rate (FPR↓) and improves robustness in OOD identification. Key contributions include: (1) a prototype-contrastive joint modeling mechanism; (2) a controllable negative embedding generation strategy; and (3) a more practical OOD evaluation paradigm. Experimental results demonstrate consistent improvements in both detection accuracy and OOD discrimination performance.

Technology Category

• Computer Vision: Object Detection & Categorization
• Search and Optimization: Learning to Search
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Neural networks that are trained on limited category samples often mispredict out-of-distribution (OOD) objects. We observe that features of the same category are more tightly clustered in feature space, while those of different categories are more dispersed. Based on this, we propose using prototype similarity for OOD detection. Drawing on widely used prototype features in few-shot learning, we introduce a novel OOD detection network structure (Proto-OOD). Proto-OOD enhances the representativeness of category prototypes using contrastive loss and detects OOD data by evaluating the similarity between input features and category prototypes. During training, Proto-OOD generates OOD samples for training the similarity module with a negative embedding generator. When Pascal VOC are used as the in-distribution dataset and MS-COCO as the OOD dataset, Proto-OOD significantly reduces the FPR (false positive rate). Moreover, considering the limitations of existing evaluation metrics, we propose a more reasonable evaluation protocol. The code will be released.

Problem

• Neural Networks
• Unseen Objects
• Recognition Accuracy

Innovation

• Proto-OOD
• Generalization Enhancement
• Evaluation Metric Innovation