Manufacturing pipelines need quality controls to ensure that products being shipped to customers are free from defects. A traditional method of quality management may involve assigning human resources to monitor product pipelines. Recent evolutions in machine learning and cloud technologies have dramatically simplified the ability to detect product anomalies. In this video, we'll take a look at one such service, called Amazon Lookout for Vision. Lookout for Vision makes it easy for anyone, even those without machine learning experience, to train custom computer vision models that detect "normal" or "anomalous" products within a manufacturing pipeline. We'll start completely from scratch and discuss gathering a fresh dataset for training and testing, training the model, improving the model, and deploying the model for production inference.