Underwater Target Detection Algorithm Based on YOLO and Swin Transformer for Sonar Images
Ruoyu Chen, Shuyue Zhan, Ying Chen
Abstract
To detect the underwater target more effectively, the paper optimizes the one-stage detection algorithm YOLO with combining Swin Transformer blocks and layers. Due to the special underwater environmental conditions, the task mainly focuses on acoustic detection methods based on sonar images. Deep learning neural network replacing the traditional detection algorithm is introduced to be the detection framework. It has been verified in the paper that the combination of lightweight network YOLO and well-performed network Swin Transformer can achieve more accurate detection precision and meanwhile meet the requirements of real-time detection using Autonomous Underwater Vehicle(AUV)’s available hardware.
Topics & Concepts
UnderwaterSonarComputer scienceTransformerArtificial intelligenceArtificial neural networkObject detectionComputer visionReal-time computingEngineeringPattern recognition (psychology)VoltageGeologyOceanographyElectrical engineeringUnderwater Vehicles and Communication SystemsUnderwater Acoustics ResearchAdvanced Neural Network Applications