To estimate AQI using smartphone camera images we require to access the device camera.
The image captured by the user is used to extract some image features on the smartphone which are sent as numbers to the server.
The image itself is not uploaded to the server, maintaining the users' privacy.
The image features sent by the user's device are fed to a Machine Learning Model. These features are also written to a CSV with date-time stamp for further training of the Model.
The extracted features are uploaded to Firebase Database and the Amazon EC 2 instance where the information is processed.
The Machine Learning Model is hosted at Firebase ML Kit.
None of the information collected by the application is sent to third party applications.