Tutorial – anomaly detection in manufacturing using AWS Lookout for Vision

In the previous section, we prepared and trained a deep learning model to classify proteins in their given categories. We went through the process of preprocessing our data, developing a model, testing the parameters, editing the architecture, and selecting a combination that maximized our metrics of interest. While this process can generally produce good results, we can sometimes utilize platform architectures such as those from AWS to automatically develop models on our behalf. Within this tutorial, we will take advantage of a tool known AWS Lookout for Vision (https://aws.amazon.com/lookout-for-vision/) to help us prepare a model capable of detecting anomalies within a dataset.

Throughout this tutorial, we will be working with a dataset consisting of images concerned with manufacturing a of Drug Product (DP). Each of the images consists of a vial whose image was captured at the end of the manufacturing...