Industrial cameras, visual photography, AI deep learning, and the style of welding joints are used to judge the quality of welding joints. Visual detection means that the incoming target is converted into an image signal through machine vision products (CCD) and transmitted to a dedicated image processing system, which is converted into a digital signal according to the pixel step by step, brightness, color and other information. The image system performs various operations on these signals to extract the characteristics of the target, and then according to the discrimination. The result is to control the equipment movement on site. The visual system has the function of deep learning. As the number of detection solder joints increases, the accuracy of the equipment will become higher and higher. Testing items include: too small solder joints, welding deviation, leakage welding, etc.