7 Frequently Asked Questions on Using Vision AI for Quality Inspection
The use of artificial intelligence is spreading rapidly in many areas, including the manufacturing floor. Manufacturers looking for ways to improve productivity and throughput, while reducing product defects and productions costs, are driving the adoption of Vision AI. Neurala is working with manufacturers around the globe to help them achieve their production metrics. Here are questions we frequently get asked:
What capabilities does vision AI have beyond computer vision?
Computer or machine vision systems are best when simple logic-based rules need to be assessed, such as orientation, part presence, measurement etc. When it comes to variability and part complexity, though, traditional computer vision often fails. Vision AI-based inspection systems can handle much more and detect anomalies such as surface-level defects, deformations, scratches, dents, etc. In addition, Vision AI can improve the throughput and productivity of the quality inspection process.
How much data/how many images do I need to get started with vision AI?
An inspection line can be set up with a very small amount of data to build the AI model. And using Neurala’s patented technology, the system learns to identify all kinds of defects. As additional data is captured, the software “learns”, and accuracy improves over time.
What if I don’t have images of every kind of defect or anomaly?
Manufacturers today have low defect rates, therefore it takes a long time and a lot of work to find images that represent the range of possible defects. With Neurala’s Anomaly Recognition capability, part of our Brain Builder software platform, anyone can train an inspection model using images of only "acceptable” parts. When the images are analyzed, and the sample contains any deviation from “acceptable”, the result is an “anomaly” recognition, even if the defect has never been seen by the model before.
Can I add Vision AI to my existing quality inspection set up?
Chances are you have a set-up for quality inspection with hardware and cameras that you know and trust. Neurala software integrates into what you already have so you can augment your vision inspection system without the expensive and time-consuming process of establishing a whole new vision system.
Do I need an AI expert on staff?
Our Brain Builder software has a simple and intuitive user interface that enables existing personnel to build Vision AI applications for quality assurance in very little time with very little data.
What are the best applications for anomaly recognition in manufacturing?
There are countless applications for anomaly recognition in the quality inspection process. For electronics, anomaly recognition can identify any aberration from the “acceptable” part and users don’t need to predict every variation of a defect; contract packagers can seamlessly switch their lines over and adjust to a new product or package products in a different way; in bottling plants, surface defects such as dented or misaligned bottle caps can easily be identified; and in metal fabrication, surface scratches or the quality of a weld can be identified as a defect.
Can I test my data or do a Proof of Concept?
With Neurala’s Brain Builder software, you can develop and test your proof of concept quickly and free of charge. Then you can be confident in your results going forward.
