IOR: Inversed Objects Replay for Incremental Object Detection

TL;DR

The paper introduces Inversed Objects Replay (IOR), a novel method for incremental object detection that reduces redundancy and improves performance without relying on generative models or labeled old-class objects.

Contribution

IOR generates old-class samples by inversing detectors, reuses objects through augmented replay, and employs high-value knowledge distillation to enhance incremental detection.

Findings

Improves detection performance in scenarios lacking old-class objects.

Reduces training and storage costs by eliminating generative models.

Demonstrates effectiveness on MS COCO 2017 dataset.

Abstract

Existing Incremental Object Detection (IOD) methods partially alleviate catastrophic forgetting when incrementally detecting new objects in real-world scenarios. However, many of these methods rely on the assumption that unlabeled old-class objects may co-occur with labeled new-class objects in the incremental data. When unlabeled old-class objects are absent, the performance of existing methods tends to degrade. The absence can be mitigated by generating old-class samples, but it incurs high costs. This paper argues that previous generation-based IOD suffers from redundancy, both in the use of generative models, which require additional training and storage, and in the overproduction of generated samples, many of which do not contribute significantly to performance improvements. To eliminate the redundancy, we propose Inversed Objects Replay (IOR). Specifically, we generate old-class samples by inversing the original detectors, thus eliminating the necessity of training and storing additional generative models. We propose augmented replay to reuse the objects in generated samples, reducing redundant generations. Moreover, we propose high-value knowledge distillation focusing on the positions of old-class objects overwhelmed by the background, which transfers the knowledge to the incremental detector. Extensive experiments conducted on MS COCO 2017 demonstrate that our method can efficiently improve detection performance in IOD scenarios with the absence of old-class objects. The code is available at https://github.com/JiaJia075/IOR.