DIODE: Dilatable Incremental Object Detection

TL;DR

DIODE introduces a dilatable incremental object detection approach that adaptively preserves old knowledge while expanding the model structure for new tasks, significantly improving multi-step incremental detection performance.

Contribution

The paper proposes a novel dilatable incremental object detector that balances old knowledge preservation and new task learning through adaptive regularization and model dilation.

Findings

Achieves up to 6.0% performance improvement over state-of-the-art methods.

Maintains high performance across multiple incremental learning steps.

Increases parameters by only 1.2% per new task for substantial gains.

Abstract

To accommodate rapid changes in the real world, the cognition system of humans is capable of continually learning concepts. On the contrary, conventional deep learning models lack this capability of preserving previously learned knowledge. When a neural network is fine-tuned to learn new tasks, its performance on previously trained tasks will significantly deteriorate. Many recent works on incremental object detection tackle this problem by introducing advanced regularization. Although these methods have shown promising results, the benefits are often short-lived after the first incremental step. Under multi-step incremental learning, the trade-off between old knowledge preserving and new task learning becomes progressively more severe. Thus, the performance of regularization-based incremental object detectors gradually decays for subsequent learning steps. In this paper, we aim to alleviate this performance decay on multi-step incremental detection tasks by proposing a dilatable incremental object detector (DIODE). For the task-shared parameters, our method adaptively penalizes the changes of important weights for previous tasks. At the same time, the structure of the model is dilated or expanded by a limited number of task-specific parameters to promote new task learning. Extensive experiments on PASCAL VOC and COCO datasets demonstrate substantial improvements over the state-of-the-art methods. Notably, compared with the state-of-the-art methods, our method achieves up to 6.0% performance improvement by increasing the number of parameters by just 1.2% for each newly learned task.