O1O: Grouping of Known Classes to Identify Unknown Objects as Odd-One-Out

TL;DR

This paper introduces a novel approach for open-world object detection by grouping known classes into superclasses and using an odd-one-out scoring mechanism, improving unknown object detection without harming known class performance.

Contribution

The paper proposes a superclass grouping strategy inspired by human cognition to enhance unknown object detection in open-world settings, addressing pseudo-label noise issues.

Findings

Significant improvement in unknown recall across benchmarks

Maintains high performance on known classes

Effective use of geometric cues alongside appearance features

Abstract

Object detection methods trained on a fixed set of known classes struggle to detect objects of unknown classes in the open-world setting. Current fixes involve adding approximate supervision with pseudo-labels corresponding to candidate locations of objects, typically obtained in a class-agnostic manner. While previous approaches mainly rely on the appearance of objects, we find that geometric cues improve unknown recall. Although additional supervision from pseudo-labels helps to detect unknown objects, it also introduces confusion for known classes. We observed a notable decline in the model's performance for detecting known objects in the presence of noisy pseudo-labels. Drawing inspiration from studies on human cognition, we propose to group known classes into superclasses. By identifying similarities between classes within a superclass, we can identify unknown classes through an odd-one-out scoring mechanism. Our experiments on open-world detection benchmarks demonstrate significant improvements in unknown recall, consistently across all tasks. Crucially, we achieve this without compromising known performance, thanks to better partitioning of the feature space with superclasses.