Peduli B3
Technologies
Overview
PeduliB3 is an undergraduate scientific writing project for a Bachelor's degree in Informatics Engineering that focuses on applying machine learning to assist the proper classification of household hazardous waste (B3). The system utilizes a fine-tuned MobileNetV2 model trained on a custom dataset consisting of household hazardous waste categories such as aerosol containers, pharmaceutical waste, used batteries, and discarded light bulbs. The application is implemented as an interactive web platform using Next.js and TensorFlow.js, featuring drag-and-drop image uploads and real-time camera access for waste classification.
The Challenge
Household hazardous waste is frequently disposed of together with general domestic waste due to limited public awareness and lack of accessible classification tools. This improper disposal leads to severe environmental issues, including soil and water contamination at landfill sites, as well as safety hazards such as explosions caused by pressurized aerosol containers mixed with regular waste.
The Solution
PeduliB3 addresses this issue by providing an accessible web-based hazardous waste classification system powered by machine learning. By leveraging a fine-tuned MobileNetV2 model, the application enables users to identify hazardous waste types through uploaded images or live camera input. This approach aims to increase public awareness and encourage proper waste separation prior to disposal, thereby reducing environmental pollution and safety risks at landfill sites.