Polluted

Polluted is an Android app that estimates the pollution level based on a real-time sky photo. It utilizes two machine learning classifiers: one for pollution level estimation and another for sky recognition. The final output is computed by incorporating additional information stored in a Firebase Firestore Database. For more details, please refer to the Wiki.

🧐 Features

Here're some of the project's best features:

• Pollution level estimation
• Firestore database geoqueries and aggregation
• Opencv image preprocessing
• Sky recognition

🧠 Classifier:

The classifiers were built using Google teachable machine. They are in tensorflow-lite format in app/src/main/assets/.

🛠️ Configuration:

After cloning the repository you have to create your own Firebase project and to update the google-services.json. For more information, please refer to the Firebase documentation.

The opencv library is already provided.

🖥️ Usage:

Wiki

😉 Info:

1. The more people use the app, the more accurate it becomes.
2. The pollution estimation classifier achieves 96% accuracy based on the considered dataset. However, it's important to note that the accuracy may be influenced by the photos taken by users.

💖Like our work?

Contact me if you have any corrections or additional features to offer me.

👥 Authors:

• Lorenzo Mirabella
• Edoardo Loni
• Marta Sanguinetti
• Alessio Vivani