Improved highway abandoned objects detection algorithm of YOLOv5

Highway abandoned object detection must be done quickly and precisely. To solve the issue of the abandoned objects detection algorithm's poor accuracy, an enhanced YOLOv5 algorithm is proposed. The Convolutional Block Attention Module is introduced to help the model recognize the abandoned objects from two dimensions. Decouple the detection heads to avoid interference between different tasks. The classification and localization tasks won't interact if the detecting head is decoupled. WIoU loss function is introduced to enhance the generalization ability of the algorithm. In comparison to the original model, the enhanced model increases the AP50 value by 4.1% to 93.9%, according to the experimental results on the self-constructed highway abandoned objects dataset. It is able to meet the needs of highway abandoned objects detection.