Glove detection system based on VGG-16 network

Abstract

Electric power enterprises play an important role in construction and development of country. Generally, the work environments of Power Grid are in high-altitude and high-voltage. In order to ensure reliable operation of national power grid and the safety of workers, workers must wear appropriate equipment to ensure safety. At present, there are few studies on workers wearing gloves. To solve this problem, we propose glove detection system based on VGG-16 network. Cameras obtain images monitored workers in real time and transmit images to the glove detection system for detecting. If it is found that workers are not wearing gloves safely, the warning is given. Because there is no dataset about gloves, we created our own dataset. Through experiments, our system achieves 84.75% accuracy, 88.67% precision and 27% false alarm rate on our dataset.