Abstract: To address the inefficiency and high false-negative rates of traditional error-proofing technologies in flexible automotive Body-in-White (BIW) production, an intelligent error-proofing system integrating the YOLOv5 deep learning algorithm with industrial controls is proposed. The system aims to achieve real-time detection and automatic correction of component misassembly or omissions, thereby advancing intelligent automotive manufacturing. A multi-module architecture is developed, encompassing system configuration, production line monitoring, feature detection, and data logging. Leveraging the YOLOv5 algorithm, the system achieves submillimeter-level recognition of critical features. A Python-PLC coordinated control mechanism enables closed-loop feedback between detection results and production line operations, while a Django-based remote maintenance platform supports dynamic model configuration updates and exception handling. Validated at a side frame assembly station in a vehicle general assembly line, the system demonstrates a zero false-negative rate and the response time for production line shutdown shall not exceed 1 s, outperforming manual inspections by 73.3% in efficiency. The system achieves rapid deployment across new stations by decoupling program logic from configuration data, offering a scalable solution for smart manufacturing. The feasibility of deep learning-industrial control integration is validated, and a technical paradigm for quality management and flexible production in automotive manufacturing is established.