Drone Detection Algorithm

This project provides the analytical overview of the data in https://huggingface.co/datasets/lll-a-p/fpv-drone-detection-dataset and provides implementation and analysis of proposed 4 algorithms: Sync Counting, FFT, Autocorrelation, Mathced Filtering. The data from dataset has to be unpacked and moved to the ./data folder.

The Project structure

• python/analysis.ipynb - Holds the dataset loader, datases analysis, methods qualitive analysis, methods implementations
• python/benchmarks.ipynb - Holds plotters for the benchmarks
• python/benchmarks_runner.py - App to collect benchmarks: connects to a board via Serial port and runs methods. Stores results in a csv file. The source passed in parameters
• src - Folder for Submodule with cross-platform firmware
• imgs - Folder with stored plots from the .ipynb files
• data - Folder for data outputs
• data/benchmarks.csv - combined results of python/benchmarks_runner.py on both ESP and STM boards.

Use

0. Pull recursively to get the Submodule with Firmware
1. Install requirements

pip install -r ./python/requirements.txt

2. Run the analysis.ipynb to get the plots from the work
3. Follow run instructions in the Submodules project with platform
4. Run the benchmarks_runner.py to obtain benchmarks
5. Run the benchmarks.ipynb to plot the results