The 360°Dynamic Vision Monitoring System for Rolling Stock is installed on the warehousing line of urban rail vehicles. The system uses image analysis technology, pattern recognition technology and deep learning technology. High-quality and high-definition images of roof, side body and bottom of various running urban rail vehicles are automatically collected, and abnormal situations such as missing and deformation of key components of roof, side body and bottom are automatically identified through image feature matching and pattern recognition technology. In the system report terminal, two-dimensional image display of visual components is automatically realized through human-computer interaction.