Repository for died cell detection with apoptosis and necroptosis distinction. It provides training and inference pipelines for training cell detection model as well as all needed data preparation steps.
Contains cell adjustor script that enables us to check parameters of cell images labeling process. Here are also all the data preprocessing/managment scripts used to prepare the dataset.
Python version tested: 3.9.*
Create environment with:
python3 -m venv <venv_name>
or this for specific version of python installed:
python3.9 -m venv <venv_name>
Activate the environment with:
source <venv_name>/bin/activate
Install requirements with:
pip install -r requirements_cell_mac.txt
or
pip install -r requirements_cell_ubuntu.txt
Yolo training pipeline with custom model wrapper and inference example.
Python version tested: 3.8.* for Mac and 3.9.* for Ubuntu.
Create environment with:
python3 -m venv <venv_name>
or this for specific version of python installed:
python3.8 -m venv <venv_name>
Activate the environment with:
source <venv_name>/bin/activate
Install requirements with:
pip install -r requirements_ultralytics_mac.txt
Setup was tested on an Ubuntu 20.04 LTS machine with NVIDIA RTX A5000 and CUDA 12.3.
Steps for your configuration may differ. Our training/inference setup requires ultralytics package with PyTorch with GPU support.
Create environment with:
python3 -m venv <venv_name>
or this for specific version of python installed:
python3.9 -m venv <venv_name>
Activate the environment with:
source <venv_name>/bin/activate
Install requirements with:
pip3 install torch==2.3.0.dev20240213+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121
pip3 install torchvision==0.18.0.dev20240213+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121
pip3 install ultralytics
