team photo

Figure 1
project photo

Computer Science and Engineering
Team 24

Team Members

Advisor

Timothy Goodwin
Juhyeon Lee
James Liebler
Emily Maciejewski

Wei Wei

sponsored by
sponsor logo

Visual Inspection Automation

Jonal Laboratories Inc. is a custom manufacturing company that primarily produces elastomeric components for the aerospace industry. Currently, Jonal uses a team of quality inspectors to manually examine each part for defects including chips, contaminations, flashes, and more. However, this is a tedious, non-value-added process subject to human error. Introducing automation would remove pressure from the inspectors, provide them more time to focus on value-added tasks, and increase the speed and efficiency of the defect detection process. Thus, this project entails creating such an automated system to classify a part image as either defective or not defective. To achieve this goal, a convolutional neural network was trained using images of both defective and non-defective parts collected onsite at Jonal in order to perform the necessary binary classification. Once an acceptable accuracy is achieved, a method to integrate the trained model into the currently inspection process, thus creating the desired automated system, is generated.