Workpiece defect detection method and device

Abstract

The invention discloses a workpiece defect detection method and device, which belong to the technical field of workpiece defect detection. The method comprises the steps that on the production line, the optical surfaces of all batches of workpieces on the production line are photographed, obtained optical surface images are input into a prediction model to be predicted, prediction data output by the prediction model are synchronously imported into a database of a simulation line, wherein the prediction data comprises workpiece batches, workpiece numbers, workpiece optical surface figure numbers and defect information; and on the simulation line, the prediction data of the to-be-optimized batch of workpieces are read from the database, and an adjustment rule corresponding to the to-be-optimized batch isformulated according to the prediction data. According to the method, a simulation scheme of the deep neural network model is adopted, on the basis that production of a production line isnot affected, computing power resources are saved, prediction results of specific batches can be conveniently, rapidly, repeatedly and stably obtained, and the working efficiency of follow-up model result optimization is improved.

Description

technical field [0001] The invention belongs to the technical field of workpiece defect detection, and relates to a workpiece defect detection method and device. Background technique [0002] In the practical process of using the deep neural network model for industrial visual appearance inspection, due to the adjustment of the workpiece production process and the replacement of abrasive tools, the prediction effect of the original model often decreases. It is necessary to optimize the results of the current forecast batch in time to meet the requirements of the production line again. In the tuning process, it is indispensable to repeatedly reproduce the process of predicting the current batch of optical surface images and returning the results. [0003] On the basis of not affecting the production line, it is an urgent problem to save computing resources and obtain the prediction results of specific batches conveniently, quickly, repeatedly and stably. Contents of the in...

AI Technical Summary

Problems solved by technology

[0002]In the practical process of using the deep neural network model for industrial visual appearance inspection, due to the adjustment of the production process of the workpiece and the replacement of the abrasive tool, the prediction effect of the original model often results. decline

Images