Garbage Classification Using Pre-Trained Models

Abstract

With the increase of urbanization, the garbage accumulated in the cities has become a big problem. While all of the accumulated garbage is classified as unnecessary, today, with the development of recycling technologies, most of the materials in the garbage are considered recyclable. Recycling recyclable materials in our daily lives is essential both for material development and for creating the life cycle of existence in the world in ecological terms. Different studies are carried out in almost every country to separate recyclable materials from the garbage. Being able to detect recyclable materials automatically with artificial intelligence will benefit from cost, human resources and time. This decomposition issue appears as a new field of study in the literature. This study aims to classify garbage by automatically using the transfer learning method. By using different transfer learning methods, it was seen that the Resnet50-V2 model showed a high success rate of 97.07% in the results.