Breadcrumb navigation

AI-based Surface Inspection for Manufacturing: NEC Advanced Analytics - RAPID machine learning [3:00]

Movie script

Narration:
We see a lot more news these days on artificial intelligence (AI) being used in business cases like these. But maybe you think there are still too many barriers stopping you from deploying AI to your company's production environment. But that's not true. By simplifying the adoption of constantly evolving AI technology, we provide effective solutions with accurate support for thinking and decision-making in business.

Torch 5 is currently a well-known open source framework for deep learning that was developed by NEC Laboratories America. RAPID machine learning has been built based on Torch 5, NEC provides fast yet lightweight software with our proprietary engine based on over twenty years of research.

Image analysis…When it comes to decision-making in business, RAPID machine learning provides high-accuracy image classification and detection to support it.

Let's look at the OneClass Classification algorithm, which is a new feature of RAPID machine learning that is extremely powerful for manufacturing inspections. Image analysis requires a huge number of pictures of both good and defective products to be collected for learning by AI, but collecting pictures of defective products is no easy matter in the precision manufacturing industry. The OneClass Classification algorithm uses proprietary NEC technology to enable defective products to be identified just by learning pictures of good products.

When a defective product is found, the defective part is highlighted to indicate it. AI-based image analysis can now optimize inspections even in the manufacturing industry, which has few defective products.

The operation screen interface also has been totally redesigned. Users can do image analysis by themselves without deep analytics knowledge according to the directions from the operation screen of RAPID.

Our RAPID machine learning is pre-tuned with fast, lightweight, high-accuracy software optimized for Intel processors. Results are achieved with one-fourteenth the computational learning complexity while keeping the same accuracy as similar engines, and generated models are a fiftieth the size. Now you can adopt AI rapidly with this single server, instead of the big server capacity and months of preparatory work previously required.
On-premises usability allows safe and simple handling of personal information like photos of faces or of high-volume videos and sensor data, without going through the cloud.

CI:
Orchestrating a brighter world
NEC

Overview

AI-based surface inspection algorithm called "One Class", which is a new feature of RAPID machine learning that is extremely powerful for manufacturing inspections. Image analysis requires a huge number of pictures of both good and defective products to be collected for learning by AI, but collecting pictures of defective products is no easy matter in the precision manufacturing industry. The "One Class" uses proprietary NEC technology to enable defective products to be identified just by learning pictures of good products. This property enables you to deploy AI rapidly to your production environment by yourself.


Related Link


Your Opinion/Comments

Did you like the contents? Please check on a scale of 5.
  • *
    Please send messages in English. Thank you for your cooperation.
  • *
    Response to comments will not be made.
  • *
    All personal information will be handled according to our privacy policy.

Contact

(July 26, 2018)