Defect Detection for Mobile Phone Cases Based on Improved Yolo Model

Abstract

As an important part of a mobile phone, the mobile phone case is the main factor affecting the appearance of the mobile phone. The mobile phone case has various kinds of defects, which seriously affect its appearance. In order to locate and classify the defects of the mobile phone case, we propose a multiscale defect detection algorithm based on artificial neural networks. The proposed model contains an anchor box generation algorithm to locate the defects using density clustering and an acceleration algorithm to boost the convolution calculation, which is critical in the production lines. We explore the effects of model parameters on the detection performance and conduct detailed experiments. Finally, we compare the proposed algorithm with traditional approaches, and observed an improvement in both detection accuracy and stability on the proposed algorithm.