HydroTrack - Python Framework for Tracking and Forecasting Clusters

Overview

HydroTrack is a Python package designed to identify, track and analyze hydrological phenomena with various data formats. Using time-varying 2D input frames along with user-specified threshold value parameters, HydroTrack is able to detect objects (clusters) and associate their displacement in time.

Algorithm Workflow

The algorithm is divided into three main modules and form the tracking workflow.

1. Feature detection: Focuses on identifying individual clusters detection from individual frame of data and extraction of features and statistics.
2. Spatial Operations: Involves spatial operations (intersection, union, difference, etc) between objects (clusters) from consecutive time steps (t and t-1).
3. Trajectory Linking: Link objects of consecutive time steps based on the spatial association.

Documentation

For a more detailed information of hydrotrack package please read the user guide available click here.

Installation

To install the Hydrotrack package, it is highly recommended to use virtual environments such as: Anaconda3, Miniconda, Mamba, or etc. And to download the package from githu you can do it using the command::

git clone https://github.com/hydrotrack-project/hydrotrack/

Create environments and install the dependencies from environment.yml file::

cd hydrotrack
conda env create -n hydrotrack --file environment.yml
conda activate hydrotrack

or install package from local file::

cd hydrotrack
pip3 install -r requirements.txt

And it is also possible to install directly from the python package repositories (pip or conda-forge)::

pip3 install hydrotrack

conda install -c conda-forge hydrotrack

Example Gallery

Open In Colab <https://colab.research.google.com/github/hydrotrack-project/hydrotrack/blob/main/examples/1_Introducing-Hydrotrack.ipynb>_.

Support and Contact

For support, email helvecio.neto@inpe.br, alan.calheiros@inpe.br