Addressing the limited recognition and detection capabilities of urban intelligent transportation systems for mixed motor and non-motor traffic flow, an object detection algorithm is proposed based on the YOLOv5 model in this study. The algorithm incorporates the GD mechanism to prevent loss of feature information between layers and enhance the algorithm’s information fusion capability while maintaining a balance between detection accuracy and speed. To extract features, the C3 module in the GD mechanism is replaced by the C3-DCNv3 module, which incorporates deformable convolution. This upgrade boosts the feature extraction capacity of the GD mechanism and enables it to effectively detect targets with more intricate shapes. A dataset is established that includes a diverse mixture of motorized and non-motorized traffic flow, upon which the detection model is trained. Experimental results demonstrate that compared to the original YOLOv5 model, the improved algorithm achieves a mAP of 74.4%, representing a 2% increase. The improved algorithm attains a detection speed of 26.32f/s, signifying a notable 18.45% boost in comparison to the original YOLOv5 model. This confirms that the enhanced algorithm is capable of basically handling the detection and recognition tasks for mixed motor and non-motor traffic flow.
Mixed Traffic Flow Detection Algorithm Based on Improved YOLOv5
Lect. Notes Electrical Eng.
International Conference on SmartRail, Traffic and Transportation Engineering ; 2024 ; Chongqing, China October 25, 2024 - October 27, 2024
Advances and Applications in SmartRail, Traffic, and Transportation Engineering ; Chapter : 3 ; 22-28
2025-07-19
7 pages
Article/Chapter (Book)
Electronic Resource
English