/blindness-detection

APTOS 2019 Blindness Detection Kaggle Competition

Primary LanguageJupyter Notebook

APTOS 2019 Blindness Detection

Setup

Add APTOS 2019 Blindness Detection dataset to data/aptos2019-blindness-detection folder at root of project.

virtualenv -p python3.6 env
source env/bin/activate
pip install -r requirements.txt
#... when done
deactivate

Training

./train.py

Submitting

TODO validate this approach and update

Inspired by: https://www.kaggle.com/c/petfinder-adoption-prediction/discussion/82092

Setting up repo

  1. Download the git repo as a zip file
  2. Upload the git repo to kaggle kernel as a dataset

Setting up the trained model

  1. Upload the trained checkpoint file to kaggle kernel as a dataset

Run

  1. Copy and paste the following into the kernel
import sys

sys.path.append("../input/path/to/root/of/blindness-detection/")

from submit import main

main("<path_to_checkpoint>", "<path_to_image_directory", "path_to_image_data_frame")
  1. Update "<path_to_checkpoint>", "<path_to_image_directory", "path_to_image_data_frame"
  2. Run kernel