STDE-YOLOv5s: a traffic sign detection algorithm based on context information enhancement
Abstract With the rapid development of autonomous driving technology, traffic sign detection has become a crucial component. Existing object detection networks encounter difficulties in accurately identifying small sized traffic signs. Considering the strong capability of the YOLOv5s algorithm in sm...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-07-01
|
| Series: | Discover Artificial Intelligence |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44163-025-00405-7 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract With the rapid development of autonomous driving technology, traffic sign detection has become a crucial component. Existing object detection networks encounter difficulties in accurately identifying small sized traffic signs. Considering the strong capability of the YOLOv5s algorithm in small object detection, its relatively low number of parameters, and ease of deployment, this paper selects YOLOv5s as the base model and proposes an improved traffic sign detection algorithm. Firstly, the feature fusion network is improved. By adding horizontal connection paths and a weighted fusion mechanism, the efficiency of multi scale feature fusion is enhanced. Additionally, by incorporating more backbone feature information into the Neck layer, the loss of detailed information is reduced, making small sized traffic signs more distinct in the images. Finally, a Context Information Enhanced Module that combines dilated convolution and deformable convolution is proposed, which enhances the network’s ability to extract context information. The proposed improved algorithm has achieved excellent results on the TT100K dataset. Its mAP50 outperforms existing advanced models such as YOLOv8 and YOLOv11. |
|---|---|
| ISSN: | 2731-0809 |