Intelligent Identification Method for Processing Target of Ceramic Green Grinding Robot

The automatic extraction of ceramic green surface processing targets (parting lines) has a significant demand in robot-based autonomous operation systems. Because traditional classification algorithms and shallow learning algorithms are not suitable for the automatic extraction of robot target information, a method for extracting ceramic green body processing targets based on deeplabv3 + is proposed. First, the image data of the ceramic green body is selected, then the deeplabv3 + model is trained using the labeled data set, and finally the recognition result of the operation target on the surface of the ceramic green body is obtained. The experimental results show that deeplabv3 + can form a good target segmentation model, and the prediction accuracy of the ceramic green parting line region of the algorithm is 87.5%. This research can provide a reference for the automatic extraction of the operation target of the ceramic green grinding robot.