ASAH (Aplikasi Sortir Sampah) is a mobile application designed to help communities manage waste more effectively. By leveraging computer vision, the app identifies different types of waste and provides guidance on proper disposal and recycling methods.
Key Features
- Real-time Classification: Instantly identifies waste types using the device's camera.
- High Accuracy: Achieved 94% accuracy in waste categorization using a fine-tuned MobileNetV2 model.
- Offline Capability: Optimized for on-device inference using TensorFlow Lite, ensuring functionality even without internet access.
- Educational Content: Provides tips and guidelines on recycling and waste management.
Technical Details
The core of the application is a deep learning model based on the MobileNetV2 architecture. We chose this architecture for its efficiency on mobile devices. The model was trained on a custom dataset of waste images and then converted to TensorFlow Lite format for deployment on Android.
The Android application was built using Kotlin, integrating the TFLite interpreter for seamless real-time performance.