kaggle-rsna-2020-9th-solution

the 9th place solution code of Kaggle RSNA STR Pulmonary Embolism Detection (https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/overview)

Solution OverView

Solution Details

Please read this post on Kaggle Discussion.

Hardware

Our team mainly use cloud machine with following spec on GCP.

OS: Ubuntu 18.04 LTS
GPU: Tesla V100 x4

Requirement

You should install docker-ce. If you don't have docker installed, please refer to this page to install it.

How to use

Data download

Plese download data to ./input from https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/data and unzip it.

build docker

$ docker build ./images -t kaggle/pytorch:rsna

Prerocess

$ sh ./bin/preprocess.sh

Stage1

# Train and extract feature
$ sh ./bin/stage_1_train.sh
$ sh ./bin/stage_1_predict_valid.sh
# Extract feature from test data
$ sh ./bin/stage_1_predict_test.sh

Stage2

# Train
$ sh ./bin/stage_2_384_train.sh
$ sh ./bin/stage_2_512_train.sh
$ sh ./bin/stage_2_concat_train.sh
# Predict test data
$ sh ./bin/stage_2_384_test.sh
$ sh ./bin/stage_2_512_test.sh
$ sh ./bin/stage_2_concat_test.sh

Stacking

# Train
$ sh ./bin/stacking_train.sh
# Predict test data
$ sh ./bin/stacking_test.sh

Blending

$ sh ./bin/blending.sh

The final output is generated in ./output/final_submission.csv.

shimacos37/kaggle-rsna-2020-9th-solution