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

TL;DR

Proto-OOD introduces a prototype similarity-based method for out-of-distribution detection, improving accuracy by leveraging feature clustering and contrastive learning, with a new evaluation protocol to better measure performance.

Contribution

It proposes a novel OOD detection network using prototype similarity and contrastive loss, along with generating OOD samples for training, and introduces a more reliable evaluation protocol.

Findings

Significantly reduces false positive rate on Pascal VOC and MS-COCO datasets.

Enhances category prototype representativeness with contrastive loss.

Proposes a new evaluation protocol for OOD detection.

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.