Emopy

Emopy is python module for training emotion recognition from face images models

Getting Started

This module contains the six types of networks for emotion recognition

How to run trainining program

python -m train --net face --emotions typeof-emotion-classsification --batch batch-size --epochs epoch-number --steps number-of-steps-per-epoch --lr learning-rate --dataset_path dataset-path --model_path path-to-save-model-without-extension --augmentation use-augmentation --verbose print-logs --sequence_length valid-only-for-rnn-networks

Where ```--net`` is type of network to train. For face image input model it is face. To use other type of network see the following list

• face : Face image input model
• dlib : dlib points features inputs model
• face+dlib : Face image and dlib points inputs model
• vgg-face : Face image inputs model fined tuned from vgg-face architecture.
• rnn : Sequential images input RNN model
• dlib-rnn : Sequential images and dlib points features inputs RNN model

--emotions

This module can be used to train either seven basic emotions(ANGER,DISGUST,FEAR,HAPPY,SAD,SURPRISE AND NEUTRAL) classifier,positive neutral emotions(POSITIVE AND NEUTRAL) classifier or positive negetive emotions classifier or positive neutral classifier values of --emotions

• all : for seven emotion classifier
• pos-neu : for positive neutral classifier
• pos-neg : for positive negative classifier

Network parameters

• batch : batch size, default 32
• epochs : number of epochs, default 100
• steps : number of steps per epoch, default 1000
• lr : learning rate , default 1e-4

--dataset_path

Path to dataset. For non sequence classifications this directory should contain train and test folders.

--model_path

Path to save model without extesion. e.g /home/user/models/model can be used to save model files json(/home/user/models/model.json) and h5(/home/user/models/model.h5).

other parameters are self explanatory.