The service can detect and report defects for quicker product resolution.
Amazon Web Services (AWS) announced this week that its Amazon Lookout for Vision service is now available for industrial use. This new cloud service is the latest addition to the AWS industrial machine learning services suite, which allows manufacturers to easily access automation solutions at a lower cost. Similar to its sibling products, Amazon Lookout for Vision leverages machine learning as well as computer vision technology to identify product defects and anomalies such as missing components, surface damage, or other similar irregularities during inspection.
Through a technique called “few-shot learning,” Amazon Lookout for Vision can develop a unique image analysis model for a customer with as few as 30 baseline images. This allows the service to accurately monitor production lines through live cameras and compare products with the images in real time with minimal training data. According to AWS, the technology can process thousands of images an hour to spot defects and anomalies. The range includes cracks, dents, incorrect colors and irregular shapes.
In addition, Amazon Lookout for Vision also reports these anomalies in real time through the “DetectAnomalies” service dashboard for quicker resolution. AWS says the technology can maintain high accuracy even with variations like “camera angles, pose, and lighting” amid shifting work environments. AWS has also made it possible for customers to share feedback on how correctly or incorrectly Amazon Lookout for Vision detects anomalies. According to AWS, this allows the technology to retrain itself when necessary to improve its analysis and ensure that it remains adaptable to different operations processes.
Amazon Lookout for Vision was primarily designed to help increase the accuracy and efficiency of quality assurance control in manufacturing environments. AWS shared how building and using computer vision models can often require copious amounts of data, as well as data science expertise, to train and deploy the technology. Besides this, changes in the production process will also result in retraining and redeploying, which can potentially set back an operation instead of improving it.
Vice president of Amazon Machine Learning for AWS, Swami Sivasubramanian shared in a press release how AWS is expecting the introduction of Lookout for Vision to assist its customers to “quickly and cost-effectively detect defects at scale to save time and money while maintaining the quality their consumers rely on—with no machine learning experience required.”
Amazon Lookout for Vision is now available in the U.S. East (N. Virginia), U.S. East (Ohio), U.S. West (Oregon), EU (Ireland and Frankfort), Asia Pacific (Tokyo and Seoul). It can be accessed via the AWS console and through AWS supporting partners with an hourly usage fee. This already includes both training the model and using the service for anomaly detection. Lookout for Vision is also compatible with AWS CloudFormation and can be used alongside other AWS suite services such as Amazon Lookout for Equipment, Amazon Monitron, and AWS Panorama. AWS will be announcing the addition of more regions in the upcoming months.
For more information visit, https://aws.amazon.com/lookout-for-vision/.
