Amazon has introduced its own deep learning artificial intelligence called Amazon Rekognition. This AI analyzes photos it is given and produces results related to them, detailing what is located in the photo as well as things related to the object. For example, it won’t just detect that a person is in the photo, but also their gender, facial expression, emotional state, what clothes they’re wearing, things they don’t have (like a beard), and more.
With Rekognition, you can locate faces within images and analyze face attributes, such as whether or not the face is smiling or the eyes are open. When analyzing an image, Rekognition will return the position and a rectangular frame for each detected face. Using Rekognition’s facial analysis, you can easily track user sentiment.
Deep learning AIs like Rekognition are initially put into a learning phase during which researchers feed them labelled images. The artificial intelligence learns what objects are — and features about them — based on these images, and in time is able to recognize them itself without any prompts. In this case, Rekognition provides a percentage of certainty about its estimations, such as being “99%” sure that a person in a photo is smiling.
Rekognition enables you to find similar faces in a large collection of images. You can create an index of faces detected in your images. Rekognition’s fast and accurate search returns faces that best match your reference face.
According to Amazon, a team of researchers built Rekognition over a span of many years; the AI is given ‘billions’ of photos every day to look at, helping refine its abilities. The AI is intended for developers; there’s a Rekognition API, as well as a Rekognition Demo available online that you can try out now. You will need an Amazon web services account to access the demo, however.
One possible use for Rekognition, says the company, is processing “millions of photos” every day, letting the AI do the work of identifying and tagging each image. The AI was designed to run at scale, per Amazon’s statement — you can have it authenticate faces to badges for large companies, or you can have it passively work as part of a security system, analyzing objects to alert if a dangerous one is identified.