Electronic products need to be inspected in the course of fabrication. Combing the technology of deep belief network（ DBN）

take the detection of fault indicators as an example and the detecting automation is realized. DBN is applied to actual system because of its advantage of unsupervised pre-training. The action state of video images of fault indicator is applied to classification experiment. The experimental results show that the DBN classification algorithm has evident advantages compared with support vector machine（ SVM） and BP neural network classification algorithm

the accuracy recognition is up to 100 %. The application of DBN algorithm in product inspection can satisfy the demand of productive measuring

reduce labor intensity of artificial tests

ease visual fatigue.