Odd-One-Out: Anomaly Detection by Comparing with Neighbors

TL;DR

This paper proposes a new anomaly detection approach that identifies scene-specific anomalies by comparing objects to their neighbors using multi-view 3D models, and introduces new benchmarks for this task.

Contribution

It introduces a novel scene-specific anomaly detection problem, leveraging multi-view 3D modeling and part-aware representations, along with new benchmarks for evaluation.

Findings

Effective detection of scene-specific anomalies

Robustness to occlusions through multi-view modeling

Benchmark datasets demonstrating method performance

Abstract

This paper introduces a novel anomaly detection (AD) problem aimed at identifying `odd-looking' objects within a scene by comparing them to other objects present. Unlike traditional AD benchmarks with fixed anomaly criteria, our task detects anomalies specific to each scene by inferring a reference group of regular objects. To address occlusions, we use multiple views of each scene as input, construct 3D object-centric models for each instance from 2D views, enhancing these models with geometrically consistent part-aware representations. Anomalous objects are then detected through cross-instance comparison. We also introduce two new benchmarks, ToysAD-8K and PartsAD-15K as testbeds for future research in this task. We provide a comprehensive analysis of our method quantitatively and qualitatively on these benchmarks.