This project developed using Efficient net-B0 architecture (model) to classify different types of trash. The model achieved accuracy of 91.95. This project enhanced our experience in the fields of image processing, deep learning, and sustainable technology applications.