Large, Complex, and Realistic Safety Clothing and Helmet Detection: Dataset and Method

Benchmark Model Rank Results
object-detection-on-sfchdYOLOv8+SCALE#1mAP@0.5:0.95: 53.3mAP@0.50: 78.6
object-detection-on-sfchdTOOD+SCALE#2mAP@0.5:0.95: 52.4mAP@0.50: 79.3
object-detection-on-sfchdTOOD#3mAP@0.5:0.95: 52.3mAP@0.50: 78.9
object-detection-on-sfchdYOLOv8#4mAP@0.5:0.95: 52.2mAP@0.50: 77.9
object-detection-on-sfchdVFNet+SCALE#5mAP@0.5:0.95: 51.4mAP@0.50: 76.6
object-detection-on-sfchdVFNet#6mAP@0.5:0.95: 51.0mAP@0.50: 76.4
object-detection-on-sfchdFaster RCNN#7mAP@0.5:0.95: 50.3mAP@0.50: 76.4
object-detection-on-sfchdFCOS#8mAP@0.5:0.95: 49.6mAP@0.50: 76.4
object-detection-on-sfchdYOLOv5#9mAP@0.5:0.95: 49.6mAP@0.50: 74.1
object-detection-on-sfchdFCOS+SCALE#10mAP@0.5:0.95: 49.5mAP@0.50: 76.3
object-detection-on-sfchdRetinaNet#11mAP@0.5:0.95: 48.9mAP@0.50: 75.9
object-detection-on-sfchdSSD#12mAP@0.5:0.95: 41.5mAP@0.50: 72.8