Setup environment:

1. I suggest using pytorch docker as a base environment

docker run nvcr.io/nvidia/pytorch:24.08-py3

2. Within the container clone the repository

git clone https://github.com/ChristofHenkel/kaggle-cryo-1st-place-segmentation
cd kaggle-cryo-1st-place-segmentation

3. And install necessary pip packages via

pip install -r requirements.txt

Required Data

download the Cryo ET competition data directly from kaggle: https://www.kaggle.com/competitions/czii-cryo-et-object-identification/data

or use the kaggle API kaggle competitions download -c czii-cryo-et-object-identification

and specify path to it in configs/common_config.py with cfg.data_folder

Training models

To train checkpoints necessary for replicating the segmentation part of the 1st place solution run training of 2x fullfits for each model. Thereby cfg.fold = -1 results in training on all data, and using fold 0 as validation. By default models are trained using bfloat16 which requires a GPU capable of that. Alternatively you can set cfg.bf16=False or overwrite as flag --bf16 False when running train.py .

python train.py -C cfg_resnet34 --fold -1
python train.py -C cfg_resnet34 --fold -1
python train.py -C cfg_resnet34_ds --fold -1
python train.py -C cfg_resnet34_ds --fold -1
python train.py -C cfg_effnetb3 --fold -1
python train.py -C cfg_effnetb3 --fold -1

Inference

Inference after models are converted with torch jit is shown in our 1st place submission kaggle kernel.

https://www.kaggle.com/code/christofhenkel/cryo-et-1st-place-solution?scriptVersionId=223259615