open_dbm/README.md

DBM Open Source Lib

---

Clone open dbm repo

• Clone repo from github

    git clone https://github.com/AiCure/open_dbm.git
  

---

Installation - Dev Mode

• Prepare a python 3 virtualenv

    pip install virtualenv (optional, only needed if it's not installed)
    virtualenv <your_env_name>
    source <your_env_name>/bin/activate
  

• Install package

    pip install -r requirements.txt
    sudo apt-get -y install ffmpeg
  

• Install Openface 2.0.6 (follow below installation steps)

    cd pkg/OpenFace
    bash download_models.sh
    bash install.sh
  

• Run script to process data

    python process_data.py --input_path abc.mp4 --output_path data/output --dbm_group acoustic facial movement
  

---

Installation - docker mode

• Build docker image

    docker build --tag dbm .
  

• Run container

    bash process_dbm.sh --input_path=/Users/vijayyadev/Desktop/input_vid --output_path=/Users/vijayyadev/Desktop/out --dbm_group="facial acoustic movement"
  

where,

    input_path: Required – Path of the file being processed. This can be either an .mp4 file or a .wav file.
    output_path: Required – Path where the raw & derived variable data will be stored.
    dbm_group: Optional – 
    Type of variables to calculate. If not specified, all variables will be calculated. The following inputs are acceptable:
    
        - facial, in which case only facial variables are calculated
        - acoustic, in which case only acoustic variables are calculated
        - movement, in which case only movement variables are calculated
    
    If the input_path was a .wav file, then only acoustic variables can be calculated. If otherwise, a warning message will be printed.
    

---

Documentation

Please add documentation link here