Eyes on the Water – Thousands of citizen photos train AI to monitor river pollution

Overview

This repository contains the final models, data, and supporting documentation for the "Eyes on the Water" project, which uses AI to monitor river health through citizen-collected photos. Wrc headlines

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/eyes-on-the-water.git
   cd eyes-on-the-water
   

2. Install the required packages:

   # You can either use the venv or conda environment
   python -m venv env
   source env/bin/activate   # On Windows use `env\Scripts\activate`
   pip install -r requirements.txt
   
   # Or create a conda environment
   conda env create -n eyes-on-the-water
   conda activate eyes-on-the-water
   pip install -r requirements.txt
   

3. Install torch & torchvision Please install the appropriate version of torch and torchvision based on your system configuration. You can find the installation instructions here.

Data

Please download the data from the following link and place it in the EOTW_data directory:

Eyes on the Water Data link

In the data directory, you will find the following folders of data wich are collected by WRC Group:

Data folder structure

└── EOTW_data/
    ├── labelled/
    │   ├── Image Classification
    │   └── Objection detection
    └── unlabelled/
        ├── ArcGIS_Dataset
        └── Big River Watch

Models

In this project, we use two models: YOLOv8 for object detection and CTran for image classification. Please download the models from the following link and place them in the EOTW_models directory:

Eyes on the Water models link

Demo

Before running the demo notebooks, please make sure you have downloaded the data and models and placed them in the appropriate directories.

Folder structure

└── Eyes-on-the-water/
    ├── Demos/
    │   └── ...
    ├── EOTW_data/
    │   └── ...
    ├── EOTW_models/
    │   └── ...
    ├── README.md
    └── requirements.txt

• Demos/Image classification.ipynb: Demo notebook for image classification using the CTran model.
• Demos/Object detection.ipynb: Demo notebook for object detection using the YOLOv8 model.