3D Deep Learning for automotive bumpers inspection
In addition to the fact that the inspection of car parts is very crucial today, users want their cars to have a smooth body. This changing demand has greatly increased production sensitivities. Therefore, automotive companies had decreased the acceptable error limit on the supplied parts. This pushes suppliers to increase their R&D projects. With this project, we aim to reduce the error rates of car bumpers to zero with computer vision and optical methods, and also to reduce the inspection cycle time for a bumper. We propose a combined 3D deep learning solution for this problem, since depth is important in defects and measurements. Moreover, we plan to avoid human error and reduce inspection times by proposing an end-to-end inspection system.