Abstract: Design of a glass cover plate defect detection device based on feature cascaded network. In response to the phenomenon that the surface quality detection process after fine cutting is mainly manual and inefficient in the current LCD glass processing process, a set of automatic glass cover plate defect detection equipment based on feature cascaded network is designed. The overall scheme, structural composition, visual system functions, and control process of the glass cover plate defect detection device design are discussed, Firstly, read the product barcode, record relevant information, and then conduct cutting accuracy measurement to detect the front and back sides of the glass. Finally, perform visual positioning detection for the cutting material. By improving mechanical structure design and visual processing algorithms, a combined algorithm of cascaded BP neural network and CNN coarse and fine classification is adopted to meet the requirements of accurate classification and detection efficiency for multiple types of defects in mobile phone glass cover plates. This algorithm is used to solve the problem of standard comparison and accurate classification of multiple types of defects in mobile phone glass cover plates, improve the detection accuracy and production efficiency of glass screens.