Arnold.
Back to projects
Data & AI

Replasc

ML-Powered Plastic Waste Awareness App

TensorFlowPythonAndroidMachine LearningComputer VisionKotlin
Replasc

About this project

Replasc (Reduce Plastic) is a capstone project that combines mobile development with machine learning to tackle the plastic pollution crisis. The app allows users to scan plastic items using their phone camera, instantly classifying the plastic type and providing guidance on proper disposal.

Beyond classification, Replasc tracks a user's recycling activity and shows the cumulative CO₂ emissions saved — turning individual actions into visible environmental impact. The ML model was trained on a custom dataset of plastic waste images, achieving solid accuracy across the seven standard plastic resin codes.

Key highlights

  • Trained a CNN-based image classification model on 7 plastic resin types
  • Integrated real-time camera inference with TensorFlow Lite on-device
  • Built CO₂ impact tracker to gamify recycling behaviour
  • Bangkit Academy capstone project in collaboration with Indosat Ooredoo Hutchison

Gallery

Replasc screenshot 2
Replasc screenshot 3
Replasc screenshot 4
Replasc screenshot 5