PlumX Metrics
Embed PlumX Metrics

Automatic Detection of Personal Protective Equipment in Construction Sites Using Metaheuristic Optimized YOLOv5

Arabian Journal for Science and Engineering, ISSN: 2191-4281, Vol: 49, Issue: 10, Page: 13519-13537
2024
  • 2
    Citations
  • 0
    Usage
  • 15
    Captures
  • 0
    Mentions
  • 0
    Social Media
Metric Options:   Counts1 Year3 Year

Metrics Details

  • Citations
    2
    • Citation Indexes
      2
  • Captures
    15

Article Description

Personal protective equipment (PPE) plays a crucial role in safeguarding the well-being of workers on construction sites by mitigating the risk of accidents and incidents. Nevertheless, incidents stemming from a lack of awareness regarding PPE usage persist. To enhance the oversight of construction laborers for accident prevention, most current approaches involve overseeing the availability and correct utilization of PPE using intricate data processing techniques. However, these methods encounter challenges related to their applicability and capacity for universal implementation. This study presents the hybridization of seahorse optimization (SHO) algorithm with the YOLOv5 model to ultimate detection accuracy. First, an improved YOLO network is constructed through scale and loss function enhancement. Second, the SHO is utilized to tune the parameters of the improved YOLOv5. A data set with 4535 construction work images was collected for training, validation and testing the hybrid SHO-YOLOv5 model. Experimental results supported by statistical tests show that the SHO-YOLOv5 is a capable method for detecting automatically the PPE with a precision value of 0.74, recall value of 0.65 and F1 score of 0.68. The SHO-YOLOv5 surpasses the benchmarked models like YOLOv8, YOLOv5, Faster-R-CNN MobilenetV3, and Faster-R-CNN Resnet50. Therefore, the proposed hybrid approach is a promising tool to support the project managers in the task of safety inspection.

Bibliographic Details

Provide Feedback

Have ideas for a new metric? Would you like to see something else here?Let us know