NunoRamos/IART-FEUP

IART-FEUP - Neural network to rate facial expressions.

Contributors

1. Maria João Mira Paulo
2. Nuno Miguel Mendes Ramos
3. Pedro Duarte da Costa

Index

1. Intruction
2. Resources
3. Project Structure

Intruction

In this project, we are going to implement a neural network for recognizing Grammatical Facial Expressions (GFEs) used in the Brazilian Sign Language. To do so we'll use ConvNetJS, a Javascript library, to train a neural network using backpropagation algorithms.

Resources

• Grammatical Facial Expressions Data Set
• Relevant Paper
• ConvNetJS - Javascript library for training Neural Networks
• Notes of the Stanford CS class CS231n: Convolutional Neural Networks for Visual Recognition

Project Structure

Convnetjs Demos

Contains a few examples of using Convuntional Neural Networks. ConvNetJS implements Deep Learning models and learning algorithms as well as nice browser-based demos, all in Javascript.

GFE Data

Pre-processed

Pre-processed data in json format.

Raw

Grammatical Facial Expressions for Brazilian Sign Language

The dataset is organized in 36 files: 18 datapoint files and 18 target files, one pair for each video which compose the dataset.The name of the file refers to each video: the letter corresponding to the user (A and B), name of grammatical facial expression and a specification (target or datapoints).

Contains:

• Datapoints files (* _ datapoints.txt): a timestamp (double) and 100 numeric attributes (double)

• Targets files (* _ targets.txt): a class attribute (interger)

Parser

Small CLI style node app to convert delimited .txt files into .json

Reports

Project reports.

webapp

Our implementation of a neural network using ConvNetJS to rate facial expressions.