Overview

This project studies how different deep learning architectures behave on Cardiac MRI classification when trained under the same general problem framing.

Models And Training

The work compares:

• MobileNetV2
• ResNet152V2
• DenseNet201
• InceptionV3

It also explores transfer learning, augmentation, class weighting, a spatial attention block, and an attention ensemble on a dataset of roughly 63,000 images.

Outputs

The project includes reproducible training flows, evaluation plots, saved checkpoints, and the accompanying paper so the results can be examined rather than simply asserted.

Why It Matters

I like this project because it sits at the boundary between applied ML experimentation and disciplined engineering: getting the training story, evaluation story, and documentation story all to line up.