A Lightweight Student Network with Dynamic Multi-Teacher Distillation for Optical Remote Sensing Object Detection

Area of Science:

• Computer Vision
• Remote Sensing
• Machine Learning

Background:

• Optical remote sensing object detection faces challenges including scale variation, slender targets, and complex backgrounds.
• Limited computational resources hinder the deployment of accurate detection models.

Purpose of the Study:

• To propose a lightweight, geometrically decoupled student network with dynamic multi-teacher distillation for improved optical remote sensing object detection.
• To enhance the localization accuracy of slender and orientation-sensitive objects with minimal computational overhead.

Main Methods:

• A lightweight student network based on YOLO11n, featuring redesigned regression branches and a geometrically decoupled intermediate branch.
• A dynamic multi-teacher distillation framework with specialized teachers for classification, regression, and topology supervision.
• Geometry-adaptive box loss and a branch-decoupled adaptive weighting strategy for knowledge transfer.

Main Results:

• The proposed model achieved a 6.8% reduction in parameters and a 13.9% decrease in GFLOPs compared to YOLO11n.
• Mean Average Precision at 50% IoU (mAP50) improved by 0.23 percentage points on the DIOR dataset.
• Demonstrated improved accuracy-efficiency trade-offs and practical inference acceleration on NWPU VHR-10 and deployment tests.

Conclusions:

• The lightweight geometrically decoupled student network effectively addresses challenges in optical remote sensing object detection.
• Dynamic multi-teacher distillation enhances localization accuracy and model efficiency.
• The proposed approach offers a practical solution for real-world remote sensing applications with limited resources.