South KoreaManufacturing2024 – ongoingLS ELECTRIC
Generated defect images for visual inspection model training
LS ELECTRIC generated defect images to improve its existing visual inspection model
Problem
LS ELECTRIC's visual inspection AI struggled with rare defects and unfamiliar fault types. These cases provided little training material for the existing model.
Approach
The company developed an AI model that generated defect images. It used the generated material to address gaps in its visual inspection system. Its chief digital officer described the work at an industrial AI forum in May 2024.
Results
No published figures
Takeaway
The company reported improved inspection accuracy without publishing the size of the gain. This project addressed training-data gaps in an existing inspection system.
Categories
- Size
- Large enterprise
- Industry
- Manufacturing
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Generative AIComputer vision
- Adoption
- Built in-house
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
CEOScoreDaily reported LS ELECTRIC executives' account of generative AI applications. The manufacturing example generated rare or unfamiliar defect images to improve an existing visual inspection model, without a numerical accuracy result.
Full details are in the original sources.