bdtools

Toolkit for reccuring tasks

Index

• Installation
• Comments

Installation

Pease select your operating system

Windows

Step 1: Download this GitHub Repository

• Click on the green <> Code button and download ZIP
• Unzip the downloaded file to a desired location

Step 2: Install Miniforge (Minimal Conda installer)

• Download and install Miniforge for your operating system
• Run the downloaded .exe file
  • Select "Add Miniforge3 to PATH environment variable"

Step 3: Setup Conda

• Open the newly installed Miniforge Prompt
• Move to the downloaded GitHub repository
• Run one of the following command:

# TensorFlow with GPU support
mamba env create -f environment_tf-gpu.yml
# TensorFlow with no GPU support 
mamba env create -f environment_tf-nogpu.yml

• Activate Conda environment:

conda activate bdtools

Your prompt should now start with (bdtools) instead of (base)

MacOS

Step 1: Download this GitHub Repository

• Click on the green <> Code button and download ZIP
• Unzip the downloaded file to a desired location

Step 2: Install Miniforge (Minimal Conda installer)

• Download and install Miniforge for your operating system
• Open your terminal
• Move to the directory containing the Miniforge installer
• Run one of the following command:

# Intel-Series
bash Miniforge3-MacOSX-x86_64.sh
# M-Series
bash Miniforge3-MacOSX-arm64.sh

Step 3: Setup Conda

• Re-open your terminal
• Move to the downloaded GitHub repository
• Run one of the following command:

# TensorFlow with GPU support
mamba env create -f environment_tf-gpu.yml
# TensorFlow with no GPU support 
mamba env create -f environment_tf-nogpu.yml

• Activate Conda environment:

conda activate bdtools

Your prompt should now start with (bdtools) instead of (base)

Comments

• Input data are modified when using get_edt with the following parameters (and maybe others):
  • get_edt(regions, target="foreground", normalize="object")
• When naming masks with suffix with Annotate, need to be able to fetch previous mask with the suffix (before and after opening the interface)
• take into consideration all image normalization issues (maybe a separate step to get more control?)
• Multi-channel input images for deep learning training
• Avoid making deep learning image check larger than GitHub limits (100mb) or just not do it at all.