Recently, Amazon announced the general availability of a machine learning (ML) service, known as Amazon Lookout for Vision, with the ability to detect defects and anomalies in visual representations using computer vision (CV).

With the help of Amazon Lookout for Vision, manufacturing companies can now increase the quality and reduce the operational costs by quickly identifying differences in images of objects at scale. The service uses ML to study images as a person would, at a higher degree of accuracy and much larger scale. The CV service comes with several features, such as dashboard view, simplified labelling, quick evaluation, and trial anomaly detection tasks and feedback.

How Lookout for Vision Works

Lookout for Vision uses machine learning techniques to see and understand images from any camera with a higher degree of accuracy at scale.

The process includes:

• Firstly, collect images from the production line and load them into the Lookout for Vision console.
• Next, label images as normal or anomalous and Lookout for Vision will automatically build a model in a few minutes. After the model is built, you can tune the machine learning model to improve defect detection by adding images to the dataset.
• You can use the Lookout for Vision dashboard to monitor defects as well as improve processes.
• The next step is to automate the visual inspection processes in real-time or in-batch and receive prior notifications when the defects are detected.
• Lastly, make continuous improvements by providing feedback on the identified product defects.

Benefits of Lookout for Vision

• Quick and Easy Improvement of Processes: The Amazon Lookout for Vision provides a faster and easy way to implement computer vision-based inspection in industrial processes at scale. It works by creating a machine learning model and analyses images from cameras that monitor the live process line to spot any differences compared with the baseline images.
• Increase in Productivity: Manufacturing companies can reduce the defects in production processes in real-time. The service identifies as well as reports visual anomalies to help manufacturers take quick action.
• Reducing the Operational Costs: Reports trends in the visual inspection data, such as identifying processes with the highest defect rate or flagging recent variations in defects. It also provides the ability to determine whether to schedule maintenance on the process line or reroute production to another machine to reduce downtimes.
• Low Setup Costs: With the help of Lookout for Vision, you can convert your less expensive cameras to ML-enabled visual inspection cameras.
• Accurate Outcomes in Challenging Conditions: Machine vision systems usually require highly controlled imaging conditions to deliver accurate results. Lookout for Vision can accurately detect defects even under different lighting conditions.
• Continuous Improvement of Accuracy: Quality managers or process engineers can continuously view and verify predicted defects using this service. Once verified, the feedback can be used to update the machine learning model, resulting in higher accuracy and improved performance.

Pricing

With this computer vision service, users can pay for what they use. Two components determine the final bill: charges for training the model and the charges for detecting anomalies. You can pay for the number of hours it takes to train the defect detection model, or you can pay for the number of hours you detect anomalies or defects with your customised model. 

As part of the AWS Free Tier, you can get started with Amazon Lookout for Vision for free. The Free Tier lasts three months and includes ten free training hours per month and four free inference hours per month.
Know more about pricing here.