YOLOOC: YOLO-based Open-Class Incremental Object Detection with Novel Class Discovery

TL;DR

This paper introduces YOLOOC, a YOLO-based detector for open-world object detection that effectively discovers and learns novel classes during inference without prior novel-class data.

Contribution

We propose a new open-world object detection framework and benchmark, enabling novel class discovery at inference time using a YOLO-based architecture.

Findings

Effective novel class discovery demonstrated on new benchmark

Label smoothing prevents over-confidence in known classes

YOLOOC outperforms existing methods in open-world detection

Abstract

Because of its use in practice, open-world object detection (OWOD) has gotten a lot of attention recently. The challenge is how can a model detect novel classes and then incrementally learn them without forgetting previously known classes. Previous approaches hinge on strongly-supervised or weakly-supervised novel-class data for novel-class detection, which may not apply to real applications. We construct a new benchmark that novel classes are only encountered at the inference stage. And we propose a new OWOD detector YOLOOC, based on the YOLO architecture yet for the Open-Class setup. We introduce label smoothing to prevent the detector from over-confidently mapping novel classes to known classes and to discover novel classes. Extensive experiments conducted on our more realistic setup demonstrate the effectiveness of our method for discovering novel classes in our new benchmark.