Computer Vision Detection of Non-conformities

This student team will work to create a vision detection system built upon a stored learning model or neural network that detects defects in painted surfaces. IDD will provide parts and examples of known defects to aid in building the learning model. Once the system is developed, this student team will work to test it on trial runs within actual production to determine validity. This student team will work to design a system with good ergonomics; the design package this student team will work to create should fit and operate on a production desk. The student team will work to design a system that can interface with an isolated network and/or database environment to store and access learning model. The vision system must also be able to detect blemishes, fod, laser burns or defects at 0.5mm in size or greater, from a viewing distance of approx. 18 inches. Additionally, the student team will work to incorporate into the system's design the capability of measuring parts, or the ability to detect when hardware, like screws, are missing. The vision system this student team will work to create must employ some recognition system whether OCR or barcode detection to recognize job and part number. The outcome this student team will work to achieve is a system that provides a PASS / FAIL display, based on image recognition, and records the results. For questionable items, it would prompt the user on screen to review. The system should be scalable for when there comes a time in the future when Safran builds additional vision detection units, Safran would just have to link it to the software/database.

Latest news