Research on Lightweight Target Detection Algorithm based on YOLOv5s

Abstract

Aiming at the problems of high model complexity and high consumption of computational resources faced by target detection algorithms when deployed on mobile and embedded devices, this study proposes a lightweight and improved algorithm based on YOLOv5s. By replacing the backbone network of YOLOv5s with MobileNetV3, the multi-scale feature expression capability is retained while reducing the number of parameters. Experiments on the COCO2017 dataset show that the improved model's number of parameters is reduced to 7.9 MB and the detection accuracy (mAP@0.5) is maintained at 84.75%. The algorithm significantly improves the computational efficiency while maintaining the detection accuracy, providing an effective technical solution for real-time target detection in resource-constrained environments.