00_Installation.md

📘 Installation

This document contains detailed instructions for installing the necessary dependencies needed to make this project work on your system.

🚧 To-Do List 🚧:

• List the required packages.
• Provide a step-by-step guide.
• Add a bash file for automatic installation.

Prerequisites:

• 🐧 The installation has been tested on an Ubuntu 20.04 system.
• 🐍 We used a (mini)Conda environment - version 4.13 - with Python version 3.10
• 🔥 At least one Nvidia-GPU (We tested on a Nvidia GTX 1080-Ti) as we use PyTorch.

⬜ Essentials

• Create and activate a conda environment:

  conda create --name READMem python=3.10
  conda activate READMem

• Install PyTorch with CUDA (our version is 12.2 [or 11.7 check again])

  conda install pytorch torchvision pytorch-cuda=12.1 -c pytorch -c nvidia

• Install panda, matplotlib, PIL etc ...

🟥 READMem-MiVOS

Download under MiVOS, the propagation branch from the official MiVOS(Mask-Propagation) repository, by following their instructions.
Replace the scripts (detailed the one), with the on in (path), through (command) / and delete the the needn't one

🟦 READMem-STCN

🚧🚧🚧🚧

🟧 READMem-QDMN

🚧🚧🚧🚧

♨️ Download datasets

We use the DAVIS 17 dataset and LV1 dataset. Download the datasets and link them to a data folder in the main project:

mkdir data
ln -s [path/to/where/DAVIS/is/downloaded] ./data/DAVIS
ln -s [path/to/where/LV1/is/downloaded] ./data/long_video_set

✅ Check if everything is alright

In the end, the repository's tree should look like this - use tree -L 1 in the main project folder:

├── MiVOS
├── READMem_API
├── data
├── dataset
├── docs
├── inference/data
├── model
├── scripts
├── util
├── .gitignore
├── LICENSE
├── README.md
├── READMem_MiVOS.py
├── TO-DO.md
├── inference_READMem_MIVOS.py
├── memory_configuration.yaml
├── requirements.txt
└── train.py