Advanced Facial Recognition with AI/ML and Graph Databases

Abstract: Graph Databases are incredibly powerful tools to quickly search data relationships. In this presentation we will combine graph technology with AI/ML to glean intelligence from a collection of photos.

Summary: The amount of visual data such as photos and videos has increased exponentially over the last decade. In the public sector especially, there is an incredible opportunity to leverage photos and videos for discovering new trends, relationships, and intelligence artifacts. In the last few years techniques such as image recognition using AI/ML have become increasingly popular and more accessible to the general public. A great example is facial image tagging, a capability that was first democratized by companies like Facebook, and is now also available from services like AWS Recognition. Services like Recognition allow individuals with no previous data science experience to simply come with a collection of images and automatically discover insights like object tagging, facial analysis, and face matching.

However, often times analysis doesn’t just stop here. In many applications, especially in the national security space, finding how one face in image may relate to others in a collection may be just as important as what is depicted in the original image. This is often a challenging problem for traditional databases and search capabilities. Unlike textual data, there are not words that can be searched on to link images in collections. This is where Graph Databases can be incredibly powerful tools, as they allow users to quickly search complex relationships within large datasets. In this presentation we will combine graph databases with AI/ML face matching technologies to create a fully automated solution to glean intelligence from a collection of photos. This solution will leverage managed services within Amazon and show how even those without a Data Science background can accomplish this advanced form of image analysis.

Presenter(s)