The Solution for the GAIIC2024 RGB-TIR object detection Challenge

TL;DR

This paper presents a lightweight YOLOv9-based solution with multi-level auxiliary branches and a feature-level fusion module for RGB-TIR object detection in UAV scenarios, addressing complex backgrounds and calibration issues.

Contribution

It introduces a novel lightweight YOLOv9-based model with multi-level auxiliary branches and a feature fusion module for improved RGB-TIR object detection in UAVs.

Findings

Achieved mAP scores of 0.516 and 0.543 on benchmarks A and B.

Maintained the highest inference speed among competing models.

Enhanced robustness to complex backgrounds and calibration issues.

Abstract

This report introduces a solution to The task of RGB-TIR object detection from the perspective of unmanned aerial vehicles. Unlike traditional object detection methods, RGB-TIR object detection aims to utilize both RGB and TIR images for complementary information during detection. The challenges of RGB-TIR object detection from the perspective of unmanned aerial vehicles include highly complex image backgrounds, frequent changes in lighting, and uncalibrated RGB-TIR image pairs. To address these challenges at the model level, we utilized a lightweight YOLOv9 model with extended multi-level auxiliary branches that enhance the model's robustness, making it more suitable for practical applications in unmanned aerial vehicle scenarios. For image fusion in RGB-TIR detection, we incorporated a fusion module into the backbone network to fuse images at the feature level, implicitly addressing calibration issues. Our proposed method achieved an mAP score of 0.516 and 0.543 on A and B benchmarks respectively while maintaining the highest inference speed among all models.