The accuracy of manual trash sorting varies greatly due to the differences in standards and management levels of countries and regions. Auto Trash Sorting System uses neural networks based image recognition system that provides >90% trash sorting accuracy.
Block diagram illustrates interconnections between systems and components
Professor Tessier is our advisor who guided us in both technical and non-technical parts in this project.
You designed image recognition system, deployed MobileNets in Vision Bonnet.
Jiang designed rotating board system, built a cabinet of the whole system.
You Zhou: youzhou@umass.edu
Jiang Chu: jchu@umass.edu