YOLOv7‐SFWC: A detection algorithm for illegal manned trucks

Abstract Automatic analysis and evidence collection of obvious traffic violations, such as illegal manned trucks, is one of the critical operational challenges of the traffic police department's business. For the enormous volume of road surveillance images generated daily, traditional manual screening is highly time‐intensive and resource‐draining. Therefore, this article proposes an improved detection model YOLOv7‐SFWC for illegally manned trucks. First of all, the pictures of illegal manned vehicles obtained by relevant departments are expanded and labeled, and the dataset of illegal manned vehicles is created. Building upon the foundational YOLOv7 model, this study replaces the traditional convolution module with the FasterNet convolution module and SCConv module, and introduces the Wise‐IoU (WIoU) loss function algorithm and Coordinate Attention (CA) mechanism. The results show that the mAP value of the YOLOv7‐SFWC model is improved by 4.15% and FPS by 7.6 compared with the original YOLOv7 model, and the computational complexity is reduced to adapt to the deployment. Moreover, the model's effectiveness is validated through extensive comparison experiments. Finally, the visual results show the accurate performance of the model and verify the progress of YOLOv7‐SFWC. This advancement has the potential to transform traffic violation enforcement by reducing reliance on manual screening, effectively combating traffic violations, and purifying traffic order.