Abstract:With the rapid development of remote sensing technology and the growing demand for applications, the classical deep learning-based object detection model is bottlenecked in processing incremental data, especially in the increasing classes of detected objects. It requires models to sequentially learn new classes of objects based on the current model, while preserving old categories-related knowledge. Existing class-incremental detection methods achieve this goal mainly by constraining the optimization trajectory in the feature of output space. However, these works neglect the case where the previously learned background is a new category to learn, resulting in performance degradation in the new category because of the conflict between remaining the background-related knowledge or updating the background-related knowledge. This paper proposes a novel class-incremental detection method incorporated with the teacher-student architecture and the selective distillation (SDCID) strategy. Specifically, it is the asymmetry architecture, i.e., the teacher network temporarily stores historical knowledge of previously learned objects, and the student network integrates historical knowledge from the teacher network with the newly learned object-related knowledge, respectively. This asymmetry architecture reveals the significance of the distinct representation of history knowledge and new knowledge in incremental detection. Furthermore, SDCID selectively masks the shared subobject of new images to learn and previously learned background, while learning new categories of images and then transfers the classification results of the student model to the background class following the judgment model of the teacher model. This manner avoids interferences between the background category-related knowledge from a teacher model and the learning of other new classes of objects. In addition, we proposed a new incremental learning evaluation metric, C-SP, to comprehensively evaluate the incremental learning stability and plasticity performance. We verified the proposed method on two object detection datasets of remote sensing images, i.e., DIOR and DOTA. The experience results in accuracy and C-SP suggest that the proposed method surpasses existing class-incremental detection methods. We further analyzed the influence of the mask component in our method and the hyper-parameters sensitive to our method.

Towards Class-incremental Object Detection with Nearest Mean of Exemplars

Incremental Learning of Object Detectors without Catastrophic Forgetting

Incremental Learning In Online Scenario

Incremental Learning of Single-stage Detectors with Mining Memory Neurons

Class-Incremental Learning: Survey and Performance Evaluation on Image Classification

Incremental Detection of Remote Sensing Objects with Feature Pyramid and Knowledge Distillation

Class incremental learning with self-supervised pre-training and prototype learning

Self-Organizing Pathway Expansion for Non-Exemplar Class-Incremental Learning

Prototypes Sampling Mechanism for Class Incremental Learning

Class-Incremental Learning Based on Big Dataset Pre-Trained Models

Learning Causal Features for Incremental Object Detection

Incremental Object Detection with CLIP

On the Exploration of Incremental Learning for Fine-grained Image Retrieval

Incremental Scene Classification Using Dual Knowledge Distillation and Classifier Discrepancy on Natural and Remote Sensing Images

Towards Non Co-occurrence Incremental Object Detection with Unlabeled In-the-Wild Data

Few-Shot Incremental Learning for Label-to-Image Translation

A Class-Incremental Detection Method of Remote Sensing Images Based on Selective Distillation

Learning to Classify With Incremental New Class

CIOD: an intelligent class-incremental object detection system with nearest mean of exemplars

Class-Incremental Few-Shot Object Detection