Research on Infrared Thermal Image Object Detection Method Based on Yolov5

To address the problem of insufficient accuracy of the traditional Yolov5 convolutional neural network model under low resolution and high background noise of infrared thermal images, this paper proposes an improved scheme by introducing the CBAM attention mechanism, the precise positioning loss function CIoU, and the generalized focal loss. The proposed approach is evaluated by conducting ablation experiments on thermal imaging datasets, and the results demonstrate an improvement in detection accuracy by 3.95% compared to other methods.