Our project's mission is to build a cross-platform mobile application compatible with IOS and Android that revolutionizes and promotes proper waste management. With our machine-learning powered application, users are able to scan their food packaging items and find an enviornmentally sustainable way of disposal.
Current applications that help identify where each type of waste goes are extremely helpful but there are some issues that we wished to solve with out application. Current applications are text-based and this means that the user would need to classify the item in order to find out what kind of waste it is. Our application lets the user identify the item usinig its barcode and then it reveals whether or not the item is Organic Waste of Recycling.
Classification Model: Enviroclean uses a trained TensorFlow model of an accuracy of about 87% obtained from a dataset of more than 25,000 household items.
Barcode Scanning: Allows accurate classification of the object as it obtains the picture of the product from the internet ensuring that the highest resolution picture is passed into the model for classfication.
Catalogue Variety: Application has a variety of household items listed in the case that no barcode is available. The user can select the item by category and the application will tell them which type of waste that household item is.
Frontend: Kivy
Backend: OpenCV, Pyzbar and BeautifulSoup
Machine Learning: TensorFlow, Keras, Google Colab
Dataset: Kaggle
Please go to the respective directory to ibtain more information about the production process and user documentation.
Android: Unable to test the application as no one in the group owns an Android device.
IOS: Unable to test the application as did not have budget and time for the application to be verfied on the App Store.
We created a demo of our application so that any user is able to seemlessly use our product whenever they want