Visualization of Emotions in App
Emote is a facial recognition SDK that works without needing OpenCV. It is beginner friendly and easily implementable. Using Computer Vision and Big Data, we collected 12 data points that are then used to analyze photos for emotions.
First we constructed a database of faceprints. This is performed by scanning in full-frontal of the people you are looking at and computing a face print for each one. The software then recognizes the face in the photo, analyzes the data from all points, compares the photo to those in the database and determines what levels of each of the following emotions are being exemplified in that image. The emotions tested include: neutral, happy, surprised, anger, fear, sadness, and disgust.
In order to prove that our application works, we created smaller Android Mobile Applications that use our SDK for more whimsical things. One application, Emoticon, analyzes the data from the 12 data points and relates them to Android based emojis. Then it displays which emoji best describes the face you're currently making. For example, if you are winking, then you will get assigned a winking emoji. We also have an android application that imports photos and then shows you the metered levels of each emotion depicted in the photo for the user to see how the backend of the SDK works.
Finally, we designed a developer's website where they can find more information on examples and how the SDK works. This site also provides a link to GitHub for easier access.