Smart Anomaly Detection and Classification for Validation Process of Watch Gears using LABVIEW

Abstract

This project aims to automate the process of validation of miniature gears such as gears that are used in watches, clocks, etc., by using machine learning algorithms. These processes are carried out manually and hence the production speed is heavily dependent on the speed of validation of humans. This project aims to withdraw human support in this area by the use of machine learning. The gears are examined by using high-resolution cameras. The output of the camera is processed using NI LabVIEW's Vision Assistant which is pre-trained using ideal and defective gears. If the input gears are in coherency with those trained 'Ideal gears', such gears are sent for assembly. If the gears are defective, a signal is sent to an actuator (say a robot) to dispose of this defective gear for recycling. This way, the project when made as an industrial machine, could increase production and decrease the Cost-tothe-Company.