kbrodt / recursion-cellular-image-classification

Recursion Cellular Image Classification

Recursion Cellular Image Classification

This is an another one approach to solve the competition from kaggle Recursion Cellular Image Classification.

44th place out of 866 (silver medal) with 0.95340 accuracy score (top 1 -- 0.99763).

Prerequisites

• NVidia apex (optional)
• rxrx1-utils

pip install -r requirements.txt

Usage

First download the train and test data from the competition link.

To train the model run

bash ./run.sh

This will generates trained models and submission file.

Approach

Detailed solution see in this presentation.