Microsoft: Recognition of emotions in photos

08 jan Microsoft: Recognition of emotions in photos

Today talk of a breakthrough that in fact not be how it has failed before, and is the Team Project at Oxford, a division of Microsoft Researchpublished the beta of a tool that recognizes the emotions of the people who appear in a photograph in particular. It is of Emotion API, the creators of this tool of Microsoft they are the same people who developed the application that recognizes the age of a person in a photo.

To use the Microsoft tool have to go to E-motion API, and upload a picture with one or more faces, or place the URL of an image uploaded to the web, and position your mouse over the rectangle that detects each face. There is displayed a range of 8 possible emotions and their level of presence in particular face.

What's behind the Emotion API of Microsoft?

The Microsoft technology behind this show combines machine learning in the Azure platformwhich allows to process a number of important information in the cloud for later integration in a given system. In the case of something like facial recognitionthe system can learn to recognize certain features of a set of images it receives, and then apply that information to identify facial features in the new images you see.

This Microsoft tool of emotion that we launched today can be used to create systems that recognize eight major emotional states: anger, contempt, fear, aversion, happiness, neutral, sadness or surprise. The tool delivers a representative percentage of the emotions that are displayed on screen. In the case of the photo you see in the header the application managed to understand the sadness in the face of the girl and gives a figure of 0.99 in this field.

The truth is that it seems surprising, and coincidentally all the photos that I upload go out with high values of happiness. Others wouldn't be an app for mobile, but let us work the boys of Microsoft Let's see if they continue to surprise us.