Dataset
| Dataset Name | Training Images | Validation Images | Class Labels | License |
|---|---|---|---|---|
| Kaggle Construction Site Safety Dataset (v3) | 2,605 | 114 | 0: Hardhat 1: Mask 2: NO-Hardhat 3: NO-Mask 4: NO-Safety Vest 5: Person 6: Safety Cone 7: Safety Vest 8: Machinery 9: Vehicle |
CC BY 4.0 |
Model Zoo
DeGirum’s PPE detection model zoo offers models trained with the Ultralytics repository.
| Model Architecture | Input Size | Precision | Runtime | Device Type | mAP 50-95 | mAP 50 | FPS |
|---|---|---|---|---|---|---|---|
| yolov8n_relu6_ppe | 640x640 | INT8 | N2X | ORCA | 0.3588 | 0.6818 | 101.1 |
| yolov8n_relu6_ppe | 640x640 | INT8 | TFLITE | EDGETPU | 0.3614 | 0.6845 | – |
| yolov8n_relu6_ppe | 640x640 | INT8 | OPENVINO | CPU | 0.3564 | 0.68 | – |
| yolov8n_relu6_ppe | 640x640 | FP32 | OPENVINO | CPU | 0.3635 | 0.6805 | – |
| yolov8s_relu6_ppe | 640x640 | INT8 | N2X | ORCA | 0.5681 | 0.8265 | 40.2 |
| yolov8s_relu6_ppe | 640x640 | INT8 | TFLITE | EDGETPU | 0.5703 | 0.828 | – |
| yolov8s_relu6_ppe | 640x640 | INT8 | OPENVINO | CPU | 0.5683 | 0.8215 | – |
| yolov8s_relu6_ppe | 640x640 | FP32 | OPENVINO | CPU | 0.5714 | 0.8272 | – |