Back to All Work
mlID22025
AgriTech Gorontalo
RicaNet — On-Device Chili Ripeness & Disease Classifier
Lightweight TensorFlow Lite computer vision model running 100% offline inside a Kotlin/Flutter mobile app to grade Cakalang & chili harvest quality in rural fields.
Technology Stack
FlutterKotlinTensorFlow LitePythonComputer Vision
Machine Learning That Actually Leaves the Jupyter Notebook
A model that only runs on an expensive cloud GPU is useless in a farm with zero signal. We trained and quantized a custom MobileNetV3 vision pipeline down to a 4.2MB TFLite model running at 30fps on budget Android phones.
Additional Screens
Primary screen displayed above.