Table 3.
Comparison of this proposal with other works.
| Work | Method | Data | Distinguished Type | Accuracy | Efficiency |
|---|---|---|---|---|---|
| Ours | Feature-based | Depth | With/Without | 96.9% | 31.55 FPS |
| Mask Type | 87.85% | ||||
| Cao et al. [14] | YOLOv4-large | RGB | With/Without | 94% | 18 FPS |
| Night Time | 77.9% | ||||
| Nagrath et al. [10] | SSDMNV2 | RGB | With/Without | 92.64% | 15.71 FPS |
| Jiang et al. [12] | SE-YOLOv3 | RGB | With/Without/ Correct Wearing |
73.7% | 15.63 FPS |
| Walia et al. [7] | YOLOv5 | RGB | With/Without | 98% | 32 FPS |
| Su et al. [11] | Transfer Learning and Effcient-Yolov3 | RGB | With/Without | 96.03% | 15 FPS |
| Mask Type | 97.84% | ||||
| Yu et al. [8] | YOLO-v4 | RGB | With/Without | 98.3% | 54.57 FPS |