Multi-modal Emotion Recognition Classification

Prerequisites

• Python 3
• Linux
• Pytorch 0.4+
• GPU + CUDA CuDNN

Dataset

CREMA-D

Pretrained Weight for Multimodal

1. Audio_Encoder
2. Visual_Encoder

Getting Started

• Installation

git clone https://github.com/PVTHust/Speech_project_Vin.git

• Download dataset:

gdown 1CdjCD2amHDsjJFfb5OuTbIfwEikgn6u-

• Unzip dataset:

unzip cremad.zip

• Train and evaluate our model:

python /content/Speech_project_Vin/main.py

Or if you use Kaggle/Jupyter notebook you can run:

train-kaggle.ipynb

and fix dataset path on config.yaml