The application aims to help users classify waste quickly and accurately, while also providing knowledge about waste classification, and creating a positive and rewarding experience for users, reinforcing their commitment to waste reduction and environmental conservation.
Watch SoWaste demo video here: SoWaste - GDSC Solution Challenge 2023
Project compose 3 git submodules:
- Mobile Application: current repository
- Backend: ./sowaste-backend
- AI Model: ./waste_detect_api
README.MD : are covered in each module for clearer explanation + installation + self-host solution
Bellow steps are required to run fully gateway application which include Flutter - Go - Python:
Download APK file here!
Requirements:
- Flutter: https://docs.flutter.dev/get-started/install
- Dart: https://dart.dev/get-dart
- Android Studio: https://developer.android.com/studio
- Xcode: https://developer.apple.com/xcode/
- Or you can use physical device to run the project. You can find the instruction here (for iOS) and here (for Android).
Steps:
- Clone project to local machine:
git clone https://github.com/GDSC-UIT/SoWaste.git- Open your terminal and run the following commands:
cd SoWaste- Run the following commands to install all dependencies:
flutter pub get- Run the following commands to run the project:
flutter run- Run the following commands in production mode:
flutter run --releaseNote: If you meet the problem from installation, you should:
- Check your internet connection;
- Run
flutter cleanandflutter run; - Delete
graddle-wrapper.jarfile fromandroid/gradle/wrapper/gradlle-wrapper.jarandflutter run.
You can find the backend installation guide here
You can find the AI model installation guide here
For our AI service, we use Yolov5 and Yolov8. TACO dataset was used to train the model. Although the model's accuracy doesn't seem to be perfect, we're still working on studies to enhance it.
Next, we use the Google Cloud Platform service to host our end-point created by FastAPI.
Here are some examples of waste our AI system has detected:
