Video analysis of hand hygiene quality of medical personnel using deep learning methods

The use of video analysis technologies based on deep learning methods is becoming increasingly popular for solving applied problems across various fields. This attention is due to the speed and quality that these technologies can provide. The experimental results presented in this paper demonstrate the suitability and effectiveness of deep learning methods for analyzing the hand hygiene procedure of medical personnel, focusing on the requirements and recommendations of the WHO. As the study shows, the task is still relevant, and the current state of the art in computer vision technology and mobile development allow for a sufficient level of quality to solve it. The key elements in the video analysis algorithm were semantic segmentation and hand actions classification. As a result, the paper presents algorithms, models, and architectural solutions selected based on computational experiments.