JEmotionRec is an application that takes an image of a person's face as input and outputs the emotion that they are expressing given a set of training data.
This project was originally developed as a project at the University of Waterloo and has since been migrated here and transformed into a more general purpose tool.
$ ./run.sh
Emotion Recognition Application
Usage: <mode> <args...>
Mode 1: fdet
Feature detection mode. When running in this mode, the args are:
arg0 = operation
arg1 = image_file (PNG or JPEG format)
arg2 = emotion_type (int)
The output format is: <file_hash>,<emotion_type>,x1,x2,...,xN
The operation can be one of full|mouth|edge, where full does face
detection, mouth applies rescaling and normalization filters,
and edge does only edge detection.
Mode 2: recognize
Recognition mode. When running in this mode, the args are:
arg0 = type (backprop or kohonen)
arg1 = num_network_nodes (int)
arg2 = num_features (int)
arg3 = the name of the training file
arg4 = the name of the testing file
The output format is: <file_hash>,<emotion_type>,p1,p2,...,pN
The following example runs the backpropagation algorithm on the given sample DRO's.
./run.sh recognize backprop 14 4 ./sample/training.dat ./sample/training.dat