AARON WARD

SURVENET - USING NEURAL NETWORKS FOR FACIAL SENTIMENT ANALYSIS
Survenet is an analysis tool built to replace certain approaches retailers undergo for measuring customer satisfaction, such as customer service emails or the HappyOrNot terminals. The application uses a trained neural network that is developed with the TensorFlow machine learning library.

The neural network was trained on a large dataset and is deployed as an API service for the proof-of-concept web application. The web application has facial detection capabilities to track a user face and take snapshots and show the their classified facial expressions on the fly, as opposed to questioning a person’s experience at a store a few weeks later.