This article uses the BP neural network decision tree algorithm to realize the steel products' surface defect and the classification with non-manual examination,extracts data basing on the flaw characteristic,and uses the BP neural network decision tree a