Emotion Recognition

Rohin Bhushan, Christie Gahm, Kyle Qian, Wenhuang Zeng

Introduction

Our project idea is to build a neural network that detects and identifies emotion on a person’s face. We will build a deep learning approach based on an attentional convolutional network, which is able to focus on important parts of the face. We also use a visualization technique which is able to find important face regions for detecting different emotions, such as angry, happy, fear, sadness, and neutral.

Data

Facial Expression Recognition 2013 (FER-2013): 35,685 48x48 pixel grayscale images of faces categorized by facial expressions of happiness, neutral, sadness, anger, surprise, disgust, and fear.

Acknowledge

Deep-Emotion: Facial Expression RecognitionUsing Attentional Convolutional Network

spatial-transformer-network