A Comparative Study of DCNN Models and Transfer Learning Effect for Sustainability Assessment: The Case of Garbage Classification

Recently, with the large development of AI, ML and DL with a wide range of different fields includes sustainability and environmental applications. Sustainability has three major pillars which are environment, economy and society in order to keep all systems balanced on earth for a larger number of generations. In this research, two modified DCNN models were implemented and tested for predicting and classifying garbage images into six types of garbage according to trashNet dataset. These models are CNN and VGG-16 and are implemented according to transfer learning aspect. Both models used in this research achieved high training accuracy on the train dataset for target classification of MSW garbage images. VGG-16 achieved higher training accuracy than CNN, 99.55% as an average, while CCN achieved 96.29% which is still high accuracy. Both models also achieved low training loss values for the same dataset details; VGG-16 got 1.20% loss compared to 12.57% for CNN.