Trash material classification is an interesting task that helps save the environment in its applications. In my project, I tackle the material classification problem using deep learning in search of the best ways to classify trash material for environmental purposes. A good convolutional model can help create useful application for sorting trash material without the need of human contact with recycling equipment. In my experiments, I try out several methods and networks, including different optimization techniques and different network architectures in order to find the best and most accurate convolutional models for the trash material classification task on benchmark datasets. I rely mostly on fine-tuning existing models in order to spend more time exploring and comparing different methods.
Use jupyter notebook
for running the training jobs.