Skip to content

INSTALL.md

Latest commit

 

History

History
88 lines (65 loc) · 3.46 KB

INSTALL.md

File metadata and controls

88 lines (65 loc) · 3.46 KB

Installation

Binaries (Windows only)

You never use Python and you are not an expert in computer science ?
No problem, you can use windows binaries to run DeepSCEM as a standalone application. The link to download the current release of DeepSCEM is in the README.md file.

To run the application, execute DeepSCEM.exe.

💡 If you want to speed up the training and prediction steps, you may want to setup a GPU if you got one.

💡 If you have somme missing .dll errors, you may need to install Microsoft Visual C++ redistributable as stated in Install TensorFlow 2 Documentation.

Run from source with Python

  • Install Python (3.8 - 3.11)
  • Create a virtual environment
    • python -m venv venv
  • Activate the virtual environment
    • windows: call venv/Scripts/activate.bat
    • linux: source venv/bin/activate
  • Check that your Python version is valid (3.8 - 3.11)
    • python --version
  • Update package installer
    • pip install -U pip setuptools wheel
  • Install requirements
    • pip install tensorflow==2.10.1 edt h5py imagecodecs PyQt5 pyqt5-tools tifffile tqdm matplotlib numpy
  • Generate *_ui.py files
    • windows convert_ui_py.bat
    • linux ./convert_ui_py.sh
  • Run DeepSCEM
    • python run.py

GPU

If you have a GPU, you may want to use it to make faster training and prediction. This guide only work for NVIDIA GPU.

Windows

First, you need to have working NVIDIA GPU and its driver installed. This is normally the case on your computer. We need to install cuda (>=11.2) and cuDNN (>=8.1) for TensorFlow 2.10.

We will install cuda using conda. You can install miniconda which is the lightest distribution for conda.

  1. Install miniconda
  2. Start conda prompt
    miniconda prompt
  3. Install CUDA and cuDNN conda install cudatoolkit==11.3.1 cudnn==8.2.1
  4. Run DeepSCEM with miniconda prompt DeepSCEM.exe or python run.py

To check that gpu is used durint training, use the task manager performance tab and check GPU CUDA and memory usage :

task-manager-gpu

To use your GPU, the only important thing, is to activate a conda environment with cudatoolkit and cudnn installed. You can install requirements in the conda environment or activate a previously created venv after starting the conda prompt.

Example of my "Windows test" setup :

image

Docker

There is no docker for DeepSCEM yet, but you may use an image with Python and TensorFlow installed to run DeepSCEM from source code.

  • TensorFlow official documentation

References