Skip to content
This repository has been archived by the owner on Jun 28, 2020. It is now read-only.

Latest commit

 

History

History

SCNN

Model and Hyper-parameters

The SCNN- xBlock model - Simple Covid Neural Network with x blocks - was designed with the objective of being simple and efficient at analyzing chest x-rays. The architecture is composed of the following:

SCNN Architecture

SCNN Model with balanced dataset, SCNN-1Block

Link to access the model weights: https://drive.google.com/drive/folders/1-3NdJBOLZVh9sbTva8srQLtdrCfLhcKY?usp=sharing

Results

Accuracy during Training

Loss during Training

Classification Report

Classification Report

Confusion Matrix

Confusion Matrix

Link to access the model weights: https://drive.google.com/drive/folders/1oVuymbP1irJu65j8bDx-sjjbMQW0V_tY?usp=sharing

SCNN Model with balanced dataset, SCNN-2Block

Results

Accuracy during Training

Loss during Training

Classification Report

Classification Report

Confusion Matrix

Confusion Matrix

Conclusion

This architecture was developed with simplicity in mind, given the not so good results on the bigger models. As such, starting really simple, we hoped the increase of CNN's blocks would improve performance to a certain point, staying at 2 blocks. Maybe with a bigger dataset, the results would be better.