HeMoDU: High-Efficiency Multi-Object Detection Algorithm for Unmanned Aerial Vehicles on Urban Roads.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Robotics

Background:

• Unmanned Aerial Vehicle (UAV)-based object detection is crucial for traffic monitoring due to flexibility and coverage.
• Increasing urban complexity necessitates advanced deep learning algorithms for UAVs.
• Existing algorithms face challenges in efficiently detecting numerous, rapidly changing road elements.

Purpose of the Study:

• To develop a high-efficiency multi-object detection algorithm for Unmanned Aerial Vehicles (UAVs).
• To address the challenge of achieving high-speed and accurate road object detection in complex urban environments.
• To improve the computational efficiency and detection accuracy of deep learning-based UAV object detection models.

Main Methods:

• Proposed the high-efficiency multi-object detection algorithm for UAVs (HeMoDU).
• Reconstructed a state-of-the-art, deep-learning-based object detection model.
• Optimized the model for enhanced computational efficiency and detection accuracy.
• Validated performance on public urban road datasets: VisDrone2019 and UA-DETRAC.

Main Results:

• The HeMoDU model demonstrated significant improvements in detection speed.
• The HeMoDU model achieved enhanced detection accuracy in urban road scenarios.
• Experimental results confirmed the effectiveness of HeMoDU on VisDrone2019 and UA-DETRAC datasets.

Conclusions:

• HeMoDU effectively improves the speed and accuracy of UAV object detection.
• The proposed algorithm is well-suited for the demands of complex urban road environments.
• This research contributes to advancing real-time traffic monitoring using UAVs.