This repository contains a final home work to be delivered at the end of the semester. The focus is a kaggle chalenge for pneumony detection in X-Ray Images.
CheXNet- Project based on https://www.github.com/jrzech/reproduce-chexnet.git and fine-tunned on the RSNA Dataset.rsna-pneumonia-retina- Project based on https://github.com/pmcheng/rsna-pneumonia.docker-utils- Contains the docker used to run the rsna-pneumonia-retina and CheXnet architectures.kernels- This directory contains a set of notebooks used do the exploratory analysis over the dataset, some extracted from the Kaggle competion.
The repository come with a README.md describing how to create a conda environment the model to run on pytorch. We improved from that loading the pre-trained weights and freezing the inner layers for the fine-tunning. In the directory we added notebooks to visualize predictions and run the retraining.
Clonned from another competitor, the code was adpated in few parts to enable the model and result analysis alongside with the training on our development environments.