IoT Hackathon - Illegal Forest Logging Detection

The Jupyter notebooks in the train folder have been based on Mike Smales' repository

Prerequisites

• Python 3.6
• Pipenv
• aws cli
• access to Slalom's sf-iot-hackathon S3 bucket

Usage

Setup

1. Create the conda environment: make setup.env
2. Download datafiles (5.6 GB): make setup.download.dataSet
3. (optional) Skip training and download models (1.5 MB): make setup.download.models

Running detection on Raspberry Pi

Install dependencies:

sudo apt-get update
sudo apt-get install libatlas-base-dev libhdf5-dev

Training the model

1. Start Jupyter notebooks: jupyter notebook
2. Run all the Notebooks in the train folder.
3. The models will be saved for further use in train/saved_models

Test the model against chainsaw sounds

1. Test one file - make test.one
2. Test all chainsaw files and print stats - make test.all

Run detection

1. make run.detect
2. Play drilling sounds. E.g. this