This is a final project done under the guidance of Dr. Wei Ding for UMB course CS470/670: Introduction to Artificial Intelligence.

Dr. Ding's website: https://www.cs.umb.edu/~ding/index.html

This project is a collaborative effort to demonstrate the importance of preprocessing and finetuning for deep learning models. There are several models used downloaded and finetuned from HuggingFace to visualize budgies and birds, which are included individually: ResNet18, ResNet152, MobileNet V2, Efficient Net, and Google Vit.

In our case, we successfully filtered and sorted our given budgie dataset and had success with ResNet 18, Mobile Net V2, and ResNet 152. Epochs ranging from 10-20 would net 90 to 98 percent accuracies. With a more powerful given GPU, accuracies would easily approach a perfect 100.