A Novel Face Super-Resolution-Based Framework for Identity Recognition of Unsafe Behavior Workers

Unsafe behavior detection applications contribute to construction safety management, among which face recognition and identity matching of the violation workers is essential to intervention: However due to the incompleteness of information in the low-resolution images, it is difficult to recognize the correct worker identity. To fill this gap, we proposed a deep face super-resolution (FSR) model-based Integrated framework jointing unsafe behavior detection with penalty management flow. By converting low-resolution face Images into high-resolution ones and adding extra structural facial information, our method enormously enhanced the recognition rate of violation workers. The proposed super-resolution-based framework has raised a novel solution to identify workers on-site and accomplished the automation workflow of safety management in construction.