Research Article
Safety Helmet Wearing Detection Based on Jetson Nano and Improved YOLOv5
Table 4
Detection performance of the models that have been improved using both the two improvement methods.
| | Models | Model format | Processing speed (ms/f) | Precision (%) | Recall rate (%) |
| | YOLOv5s-SN | pt | 44.9 | 87.7 | 79.2 | | YOLOv5s-SN (after optimization) | Engine | 32.2 | 88.8 | 78.0 | | YOLOv5m-SN | pt | 64.6 | 88.8 | 82.6 | | YOLOv5m-SN (after optimization) | Engine | 56.5 | 88.0 | 83.2 | | YOLOv5l-SN | pt | 93.1 | 89.3 | 84.3 | | YOLOv5l-SN (after optimization) | Engine | 83.9 | 88.6 | 84.7 |
|
|
