JapanManufacturing2015 – ongoingNippon Steel
Deep-learning detection of displaced coil cushioning
Nippon Steel detected about 80% of coil-cushioning problems with crane-mounted cameras
Problem
Protective cushioning could shift or fall while cranes moved steel coils, exposing products to scratches and dents. Different lighting conditions and coil sizes made conventional image processing difficult.
Approach
Nippon Steel worked with NEC to use RAPID deep learning at its Nagoya works after initial discussions in June 2015. The team generated altered training images to represent cushioning failures. Cameras on cranes sent images wirelessly to a central server for detection.
Results
Takeaway
False detections mattered because unnecessary line stops reduced productivity. The system balanced detection coverage against those interruptions.
Categories
- Size
- Large enterprise
- Industry
- Manufacturing
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Partner co-development
- Evidence
- Vendor case study
Sources
Source summaries
NEC described the deployed camera inspection system at Nagoya works. It reported detection of about 80% of cushioning problems, with false detections below 1%.
Full details are in the original sources.