Incremental-DETR: Incremental Few-Shot Object Detection via Self-Supervised Learning

TL;DR

This paper introduces Incremental-DETR, a method for incremental few-shot object detection that combines fine-tuning, self-supervised learning, and knowledge distillation to detect new classes with minimal data while retaining knowledge of existing classes.

Contribution

It proposes a novel incremental few-shot detection approach using DETR, incorporating self-supervised pseudo-labeling and knowledge distillation to improve performance and scalability.

Findings

Outperforms state-of-the-art methods significantly

Effective in detecting novel classes with few samples

Maintains performance on base classes

Abstract

Incremental few-shot object detection aims at detecting novel classes without forgetting knowledge of the base classes with only a few labeled training data from the novel classes. Most related prior works are on incremental object detection that rely on the availability of abundant training samples per novel class that substantially limits the scalability to real-world setting where novel data can be scarce. In this paper, we propose the Incremental-DETR that does incremental few-shot object detection via fine-tuning and self-supervised learning on the DETR object detector. To alleviate severe over-fitting with few novel class data, we first fine-tune the class-specific components of DETR with self-supervision from additional object proposals generated using Selective Search as pseudo labels. We further introduce an incremental few-shot fine-tuning strategy with knowledge distillation on the class-specific components of DETR to encourage the network in detecting novel classes without forgetting the base classes. Extensive experiments conducted on standard incremental object detection and incremental few-shot object detection settings show that our approach significantly outperforms state-of-the-art methods by a large margin.