Damowerko / mtt

Multi Target Tracking

Multi target tracking with a fully convolutional encoder-decoder.

Installation

1. Install (poetry 1.12)[https://python-poetry.org/docs/master/#installing-with-the-official-installer]. You can use the official installer or the pipx installer, whichever you prefer. If you use pipx you can install poetry using the following command.

pipx install git+https://github.com/python-poetry/poetry.git

2. By default poetry will automatically create a virtualenv for this project. If you do not want this behaviour run the following. I recommend pyenv and/or conda for managing your environments.

poetry config virtualenvs.create false --local

3. poetry install and all pip packages will install. This might take a while (up to ~10 minutes). If it's taking long it might also be a networking issue (sometimes pypi.org resolves to an IPv6 address, which may cause problems, try seeing if pip install numpy fails).

Models

Model weights are available on my Google Drive, since they are too large for git-lfs. Download the contents of the models directory and place them in the models directory.

Alternatively, you can download them using gdown.

gdown --folder --id 17jW5YQpbT6i6gV2smHl69ezC5P8VhjAj
mv mtt-models/ models/