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.

Instance-Level Scaling and Dynamic Margin-Alignment Knowledge Distillation for Remote Sensing Image Scene Classification

Pair-Wise Similarity Knowledge Distillation for RSI Scene Classification

Dynamic Convolution Covariance Network Using Multi-Scale Feature Fusion for Remote Sensing Scene Image Classification

Lightweight High-Resolution Remote Sensing Scene Classification via Adaptive Enhanced Knowledge Distillation.

TAKD: Target-Aware Knowledge Distillation for Remote Sensing Scene Classification

Lightweight Remote Sensing Scene Classification Based on Knowledge Distillation

From Coarse to Fine: Knowledge Distillation for Remote Sensing Scene Classification

Instance-Aware Distillation for Efficient Object Detection in Remote Sensing Images

Multispectral Scene Classification via Cross-Modal Knowledge Distillation

Remote Sensing Image Scene Classification with Noisy Label Distillation

Tolerant Self-Distillation for Image Classification

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

Class-Incremental Learning for Remote Sensing Images Based on Knowledge Distillation

Remote Sensing Scene Classification Using Sparse Representation-Based Framework with Deep Feature Fusion

Pixel Distillation: A New Knowledge Distillation Scheme for Low-Resolution Image Recognition

Information Fusion with Knowledge Distillation for Fine-grained Remote Sensing Object Detection

RS-SSKD: Self-Supervision Equipped with Knowledge Distillation for Few-Shot Remote Sensing Scene Classification

SIGKD: A Structured Instance Graph Distillation Method for Efficient Object Detection in Remote Sensing Images

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

Consistency- and dependence-guided knowledge distillation for object detection in remote sensing images

A class distribution learning method for few-shot remote sensing scene classification