最終更新日時： 2021/07/30 15:25:20
Automating visual inspection with image inspection based on visualization and artificial intelligence
We have realized visual sensory inspection that had previously been difficult. Our MINESPECT deep learning appearance inspection device has been proven in more than two years of actual factory operation.
- Business areas and strengths
We have succeeded in automating appearance inspections through machine learning such as deep learning. This was done with technologies like developing lighting systems for visualizing objects that are difficult to visualize although they can be visually identified, and objects invisible to the human eye, designing and developing imaging devices, and designing algorithms for performing color space operations on acquired image data and narrowing it down to the target characteristic dots. Our MINESPECT DeepLearning inspection system already has a track record of being successfully used at manufacturing sites as factory equipment for more than two years.
- Overview of main products and technologies
Taking full advantage of visualization imagery and artificial intelligence has enabled automation of previously difficult visual sensory inspections. Our MINESPECT DeepLearning image inspection device is an appearance inspection system based on artificial intelligence. MINESPECT even automates the inspection of products that had been inspected by human eyes because of variation produced in normal conditions during their manufacturing process. For example, even where "craftsman's intuition" had been used to judge defects because the defective state is very similar to the normal state, a photo-taking pod equipped with lighting that simulates human vision is installed on the production line. The acquired images are processed using proprietary image processing to reproduce inspection standards comparable to visual inspection, and inspection results are produced on a real-time basis.
- Kanto Head Office, SMRJ