)
)
)
How can we help you?
)
Marija Mijic
HR & Office Allrounder
T +43 660 383 95 46
office@craftworks.at
craftworks GmbH
Schottenfeldgasse 20/6a
1070 Vienna
Diminished Pseudo Scrap
through Computer Vision
Our innovation enhances Miba's capacity control, streamlining operations, and optimizing resource utilization.
:quality(80))
Success Story
reading time: 3 min
Automotive
Manufacturing
Visual Inspection
Industrial Software
89%
Accuracy in visual inspection
Reduced
Pseudo scrap
Less waste
Through automation and predictive maintenance
Context
Miba is a large scale supplier of components for the automotive industry. Due to the high demands placed upon automotive components, Miba has to adhere to strict quality standards for their products. Quality inspection was conducted via cameras. However, the cameras were classifying too many parts as faulty, so a more accurate system was required.
Challenge
To differentiate between actual scrap components and components that met the quality demands, all done in real time.
Assignment
craftworks was tasked with developing a computer vision processing technique to distinguish between actual scrap parts and viable components for shipment.
Solution
A deep learning model was trained and evaluated as well as deployed as a Docker service, for tests on-site. A revolutionary method of combining images from two cameras and processing the imagery lead to an above average accuracy. The results from the on-site tests were analyzed and what we discovered was that a more precise detection was taking place during the recalculation of the imagery. Technologies used included Keras, OpenCV and Docker.
With the help of advanced computer vision technology, Miba significantly diminishes pseudo scrap, revolutionizing planning, capacity control, and storage management for enhanced operational efficiency.
:quality(80))