Lightweight UAV Remote Sensing Image Vehicle Detection Method Based on Improved YOLOv5s
This paper proposes a lightweight detection model based on YOLOv5s for the real-time detection of ground vehicle targets under the condition of UAV airborne hardware. The model introduces a grouped convolution kernels into the backbone network for fea-ture extraction to reduce the number of operational parameters of the backbone network,and at the same time,an efficient attention mechanism is set at the end of each feature extraction layer to improve the weighting and screening of positive sample features. A multi-scale feature fusion layer is set in front of the feature enhancement end to further improve the information richness in the output fea-ture map. The experimental results show that the proposed improved model is superior to the original model in terms of detection accu-racy,speed and model volume. It can show good generalization ability in various complex scenes such as night scenes and highways and can be deployed in UAV airborne hardware to carry out real-time detection of vehicle objects.
万方数据
维普
