OCT-eyeDiseaseClassification

SJTU Digital Image Process

This project uses current models (resnet and vgg) to do OCT eye-disease classification

Its results can be used as Baseline of OCT research

# quick start
pip install -r requirement.txt

python main.py -h
python Eval.py -h

Type these two commands in your command line to see help menu for this project.

Also you can use sklearn.metric to do your own metric measurement.

To complete this project, you may need to add folder ./results/{model_name}/ and ./img/{model_name}/ to your local position.

The data set link is https://jbox.sjtu.edu.cn/l/SFh0f6

---

Sample image:

---

New section: Neural Architecture Search (./nas)

Reference:

"Neural Architecture Search without Training": https://github.com/gmh14/RobNets

"When NAS Meets Robustness: In Search of Robust Architectures against Adversarial Attacks": https://github.com/BayesWatch/nas-without-training