debayanmitra1993-data / blindness-detection-diabetic-retinopathy-

This is a 5 Class Image Classification Task based on a Kaggle dataset from Eye Images (Aravind Eye hospital) - APTOS 2019 Challenge. The goal is to predict the Blindness Stage (0-4) class from the Eye retina Image using Deep Learning Models (CNN). This Automated System would speed up Blindness detection on Patients

This case study is based on a Kaggle dataset - https://www.kaggle.com/c/aptos2019-blindness-detection/

Here is the detailed blog - https://medium.com/@debayanmitra1993/blindness-detection-diabetic-retinopathy-using-deep-learning-on-eye-retina-images-baf20fcf409e

Here is the arXiv.org Research paper that was attempted to Implement - https://arxiv.org/pdf/2003.02261.pdf

This Research paper is summarized in the Research paper (summary doc).pdf

Contents of the Code Files are given below :-

S No	Section	Notebook (.ipynb)	Description
1	Exploratory Data Analysis	1_eda.ipynb	All Data Analysis & Insights on Images and classes
2	Image processing, Train/validation data	2_process.ipynb	Image Resizing,preprocessing,data splitting
3	Transfer Learning using ResNet50	3_resnet50(colab).ipynb	Complete Model and Evaluation
4	arXiv.org Research Paper Implementation	research_paper_implementation.ipynb	Complete Summary of the reseach paper in Research paper (summary doc).pdf, Attempt to replicate as close as possible

blindness-detection-diabetic-retinopathy-'s Issues

Trained model

hey @debayanmitra1993-data,
Can you upload the trained model and a few example images (each class) to the repo? It would be really helpful...

Thank you,
Avinash

What is the point of this code?

i was getting error ( FileNotFoundError: [Errno 2] No such file or directory: 'df_train_train' ). please help me with this

i didn't understand about df_train_train and df_train_test

i was getting error ( FileNotFoundError: [Errno 2] No such file or directory: 'df_train_train' ). please help me with this

I didn't understand the term df_train_test

Accuracy with other models

Hey @debayanmitra1993-data , amazing notebook! Really helpful. However I was curious if using a Sequential model would be optimal here? What precaution do I need to take to reproduce the accuracy in that case?