A surface defect non-destructive inspection software has been developed for the manual inspection of notebook surface quality in the electronics industry. Initially, the surface of the notebook shell is roughly positioned by the fixture mechanism. Through the feedback of motor A and B phase encoding, hard triggering of an 8K line scan camera and a high-frequency light source controller are used to achieve time-division strobing. The complete image obtained is divided into multiple equal-sized pictures according to the customer's indicated point position map. Using the JSON HTTP protocol, the algorithm server is requested to obtain the inference result of a single image. Finally, the product surface defect types are obtained through sample-level result queries.