This system can solve the drawbacks of manual detection and instrument detection, to achieve high-speed, accurate real-time online PCB defect detection, improve the output of PCB board and the productivity of the electronics manufacturing industry, divided into hardware and software two parts. The main modules of the overall design function of the hardware part include: the core controller of the image acquisition device, the image sensor, the wireless transmission module and the manipulator. The overall functional design of the software part mainly includes the following three modules, image preprocessing module, PCB defect detection and recognition algorithm based on deep learning and background content analysis management system. The system can finally accurately locate the defective part of PCB, classify and analyze the detected defects, improve the production line and improve the yield of products. The interface of PCB defect detection system based on machine vision and deep learning is simple and clear, and the operation is easy to use. The system flow is as follows: the PCB board is placed under the image sensor by the manipulator, the strawberry pie with Raspberry Pi OS is the main control in the hardware control system, communicates with the image sensor module, transmits the image data to the defect detection system on the PC side for image analysis and detection, and transmits the detection result back to the main control, and the MCU on the manipulator communicates after the main control receives the signal, then through the MCU control manipulator on the steering gear to grab the PCB board quality classification, PCB board defect detection results can also be displayed on the display screen.