Skip to content

Latest commit

 

History

History
160 lines (104 loc) · 5.46 KB

INSTALL.md

File metadata and controls

160 lines (104 loc) · 5.46 KB
Raw

Installation - On System

This section has instructions to install the necessary tools and libraries to run GREEM directly on the system.

Prerequisites

This section contains the required libraries for GREEM.

Python 3

Python 3 is required to be installed on the system. To test if Python 3 is installed on the system, enter the command:

# check if Python is installed
python3 --version

This command should prompt the Python 3 version installed on the system.

If Python 3 is not installed, use the following command:

# install Python 3 on the system
sudo apt-get update && sudo apt-get install python3 -y

Anaconda/Miniconda

To install the Python virtual environment with all required libraries, GREEM uses Anaconda/Miniconda environments.

Install Miniconda from CLI

If Anaconda/Miniconda is not installed on the system, the following commands provide a quick installation guide to install Miniconda.

mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm -rf ~/miniconda3/miniconda.sh
~/miniconda3/bin/conda init bash

Miniconda is a minimal installer for conda that only includes the conda binaries, Python and the packages they depend on.

NVIDIA & CUDA Drivers

In order to be able to utilize NVIDIA GPUs for the benchmarks, ensure that NVIDIA and CUDA drivers are installed on the system.

FFmpeg

To use the video processing benchmarks in the encoding and decoding folders, codec libraries need to be installed first.

To test if FFMPEG is properly installed by executing the command:

ffmpeg -version

Install FFmpeg Video Codec Libraries

A command including many codec libraries is shown below.

# update aptitude
sudo apt-get update -y

# install FFmpeg libraries
sudo apt-get install -yq  libaom-dev libass-dev libc6 libc6-dev libfreetype6-dev \ 
                          libgpac-dev libmp3lame-dev libnuma-dev libnuma1 libopus-dev \
                          libsdl1.2-dev libsdl2-dev libtheora-dev libunistring-dev \
                          libva-dev libvdpau-dev libvorbis-dev libvpx-dev \
                          libx264-dev libx265-dev libxcb-shm0-dev libxcb-xfixes0-dev \
                          libxcb1-dev zlib1g-dev

FFmpeg with NVIDIA GPU Support

The guide How to install FFmpeg with NVIDIA GPU support lists the necessary steps to install FFmpeg with NVIDIA GPU support.

An alternative guide how to do so can be found at NVIDIA docs.

This step is required if you plan to process videos with GPU support.

Install FFmpeg

Use the following command to install FFmpeg if it is not already installed.

# Install FFmpeg
sudo apt-get install ffmpeg -y

Setting up the Python Environment

The Python dependencies can be installed using Anaconda with help of the environment.yml file that is located at the root folder of this repository.

Create Anaconda Environment

The environment.yml at the root level is used to create a Python environment called greem. To do so, use the command:

# create anaconda environment
conda env create -f environment.yml

This will install the Anaconda environment that is used for GREEM.

Once the environment is installed, activate it with

# activate anaconda environment
conda activate greem

After successfully installing the Anaconda environment, it is required to locally install GREEM using the command

# setup Python GREEM paths
pip install -e .

at the root level of this repository. This will install the projects as a module and setup the paths that can then be easily important inside the project itself.

Video Processing Benchmarks

In order to use the encoding/decoding benchmark Python scripts, it is required to follow the installation steps below.

Download and Add Videos to Benchmarks

Within the benchmark folder, two Python scripts can be found to download video files.

The download_full_input_files.py script can be used to download input videos from Youtube.

The download_segments.py script downloads 500 video segments with a length of 4 seconds each from the Alpen-Adria University servers.

(Optional) Installing Prerequisites on AWS Instance

How to install FFMPEG with NVIDIA GPU Acceleration on Linux