This repo contains a collection of feature extractors for multimodal emotion recognition.

Clone this repository:

$ git clone --recurse-submodules https://github.com/gangeshwark/multimodal_feature_extractors.git

1. Install FFMPEG and OpenCV from source.
2. Install the packages as specified in requirements.txt

Currently, these modalities are covered:

1. Video

This feature extractor contains uses Openface to extract and align faces and uses Face VGG to extract facial features from every frame.

Module: from src.video.models import OpenFace_VGG in your data processing code.

• Video feature extractor.
• Add text feature extractor.
• Add audio feature extractor.
• Code cleanup.

1. Soujanya Poria for his invaluable inputs.
2. Authors of caffe-tensorflow, openface