Out-of-Distribution Object Detection in Street Scenes via Synthetic Outlier Exposure and Transfer Learning

TL;DR

This paper introduces SynOE-OD, a novel framework that uses synthetic outlier data and transfer learning to improve out-of-distribution object detection in street scenes, achieving state-of-the-art results.

Contribution

The work presents a unified detection framework leveraging generative models and open-vocabulary detectors to enhance OOD detection in street scenes, addressing limitations of existing methods.

Findings

Achieves state-of-the-art average precision on OOD detection benchmarks.

Utilizes synthetic outlier data generated by models like Stable Diffusion.

Enhances robustness of object detectors to out-of-distribution objects.

Abstract

Out-of-distribution (OOD) object detection is an important yet underexplored task. A reliable object detector should be able to handle OOD objects by localizing and correctly classifying them as OOD. However, a critical issue arises when such atypical objects are completely missed by the object detector and incorrectly treated as background. Existing OOD detection approaches in object detection often rely on complex architectures or auxiliary branches and typically do not provide a framework that treats in-distribution (ID) and OOD in a unified way. In this work, we address these limitations by enabling a single detector to detect OOD objects, that are otherwise silently overlooked, alongside ID objects. We present \textbf{SynOE-OD}, a \textbf{Syn}thetic \textbf{O}utlier-\textbf{E}xposure-based \textbf{O}bject \textbf{D}etection framework, that leverages strong generative models, like Stable Diffusion, and Open-Vocabulary Object Detectors (OVODs) to generate semantically meaningful, object-level data that serve as outliers during training. The generated data is used for transfer-learning to establish strong ID task performance and supplement detection models with OOD object detection robustness. Our approach achieves state-of-the-art average precision on an established OOD object detection benchmark, where OVODs, such as GroundingDINO, show limited zero-shot performance in detecting OOD objects in street-scenes.