Standardizing and Early Warning of Sewing Beginners’ Posture Based on CNN Visual Recognition Technology

This study focuses on the posture problem in the sewing process and utilizes computer vision recognition technology to identify and warn the posture status of sewing workers, thereby reducing the risks and stress during the sewing process. By analyzing actual sewing scenarios, conducting questionnaires, summarizing expert opinions, and categorizing different sewing postures, the specific requirements and technical roadmap of computer vision recognition in sewing conditions are determined. OpenPose is used to extract the posture features of novice sewers in the sewing process, and then a deep learning network is built and trained using the PyTorch framework, enabling the model to provide early warnings for incorrect sewing postures of novice sewers.