This project is part of the Li4Lam workshop held in Addis Ababa between October 27-30, 2024.

Project Setup

This is a Python 3.9 project. First create a virtual environment in this projects directory and install the required packages (this might take some time):

python -m venv venv

# In cmd.exe
venv\Scripts\activate.bat

# In PowerShell
venv\Scripts\Activate.ps1

# install provided packages
pip install -r requirements.txt

Data access

Download the necessary data (4GB) here (account is required): UAV dataset

Or get the data via USB stick from one of the presenters.

Training

To train a model, execute the script main.py. Please take care to change the BASE_PATH variable according to your directory structure.

To run inference on your trained model execute the inference.py script. It is necessary to change your filepaths accordingly as well.

Cloud solution

The file ethiopia_uav.ipynb contains the whole functionalities of this repository in a .ipynb notebook. Upload this notebook into a Cloud IDE such as Google Collab.

Sources

Data: http://dronedataset.icg.tugraz.at

Code (modified): kaggle