Neurala Introduces Anomaly Recognition

Now enabling manufacturers to quickly configure custom visual AI models to classify acceptable and defective objects on the production line

We are excited to share a new software approach that allows manufacturers to deploy visual systems in minutes for defect or quality control inspection. With this capability, manufacturers will be able to set up or re-configure production line in a flexible way to cope with changed market or production requirements.

Our new software release leverages our proprietary anomaly recognition capabilities to enable manufacturers to set up an AI-powered visual quality control system even in those cases where it is not possible to collect large samples of defective product, a common issue in the industry, and a problem not addressable with traditional machine vision or AI.

Our anomaly recognition technology overcomes this common barrier in adopting vision AI in production environments, enabling manufacturers to streamline their inspection systems. With anomaly recognition, we can identify any product that deviates from the “acceptable” images that the customer trains its system on. Anomaly recognition allows users to:

  • Implement a vision system without defining hundreds of features to “catch” defects, as is necessary with computer vision
  • Overcome imbalanced datasets
  • Easily identify defective products

Anomaly recognition allows manufacturers to save time, plug into the hardware they already have available, and does not require in-house expertise. Users can set up a new job or modify an existing process in a fraction of the time, making production lines adaptable to changing requirements and product output. While many AI integrations require in-house technical capabilities like script writing, AI expertise, or branch logic, our Brain Builder platform has an easy, user-friendly interface that allows quality control managers and vision specialists alike to deploy vision AI for inspection. Unlike traditional computer vision, you don’t need to predict every way a defect could present itself.

“In manufacturing, having representative images of defective product is rare,” says Neurala CEO and co-founder Max Versace. “Identifying defective product is critical, but high-yield manufacturers are unlikely to produce defective items at a rate that would provide the required data. This advancement allows manufacturers to be fast, flexible, yet high-quality in a world where changing market or product requirements call for agility and configurability of production lines.”

With inspection customers like IHI, Optelos, Avisight, and Badger already using our technology, we have developed deeper insight into the problems facing industrial manufacturers today. This new capability allows manufacturers, including packaging and assembly companies, to quickly set up and modify their production lines, maximizing their ability to service a world where demand and requirements may fluctuate frequently. While anomaly recognition has already been made available to some partners, we are so excited for this capability to become widely available today.

For more information about anomaly recognition, request a demo.