Road target detection methods, detection devices, electronic equipment and storage media
By combining the BiSeNet and YOLO models, and integrating multi-layer feature information fusion and iterative training, the problem of low detection accuracy of existing models in complex traffic scenarios is solved, and higher accuracy road target detection is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 苏州万集车联网技术有限公司
- Filing Date
- 2022-11-03
- Publication Date
- 2026-05-26
AI Technical Summary
Existing Faster R-CNN, Mask R-CNN, SSD network, and YOLO object detection models cannot effectively learn feature information in complex traffic scenarios, resulting in low accuracy in road object detection.
An initial road target detection model is constructed by combining the BiSeNet semantic segmentation model and the YOLO target detection model. By fusing the short-term dense cascaded (STDC) structure, the CSP structure and SPP structure of the YOLO target detection model, and the shortcut structure, multi-layer feature information is fused. The target road target detection model is obtained through iterative training using the SIOU loss function and backpropagation algorithm.
It improves the accuracy of road target detection, and can better integrate multi-layer feature information to enhance detection results.
Smart Images
Figure CN115775366B_ABST