Waste Material Detection

• Author(s) / Creator(s)

  • Kotecha, Twisha

• We use plastic every day. Plastics can be found anywhere, in households and businesses. Plastic dependency has fueled the global production of plastic resins and fibers, growing at an annual compound rate of 8.4 percent. As a result, strong waste management is critical. A traditional waste management system is not so efficient and handy for regular households or small businesses. People do not recycle their waste correctly. Smartphones and apps can be used to overcome this issue. Such apps can perform real-time material detection and allow better waste management. The goal of this project is to create a material detection app that can perform smart waste management for household and small-scale businesses. The Android app analyses a live camera feed and recognizes materials of detected objects in real-time using a machine learning model which is based on TensorFlow Lite and TensorFlow. Material detection and waste classification are carried out in the TensorFlow framework using a pre-trained material detection model. TensorFlow Lite allows running TensorFlow machine learning (ML) models in Android apps. The TensorFlow Lite system renders prebuilt and customizable execution environments for running models on Android quickly and efficiently, along with the hardware acceleration option. We performed experimental analysis on various pre-trained models to analyze their accuracy. The developed app works for a single object at a time. It works in an indoor environment with good lighting conditions and requires a clear background.

• Date created

  2022-01-01

• Subjects / Keywords

  • Material Detection
  • Deep Learning
  • TensorFlow
  • Machine Learning

• Type of Item

  Research Material

• DOI

  https://doi.org/10.7939/r3-wb8a-6w63

• License

  Attribution-NonCommercial 4.0 International