EIP

Edge Image Process.

Install From Git

Install directly from the official GitHub repository:

pip install eip-cli

Upgrade to the latest main:

pip install --upgrade --force-reinstall eip-cli

Verify The Install

After installation, verify that the CLI and TUI both start:

eip --help
eip --tui

Run the current smoke-path workflow with a pre-segmented case:

eip run --input /path/to/case.image.nrrd --seg-input /path/to/case.seg.nrrd --backend none --output /path/to/out

Deep Learning Runtime

torch and nnUNetv2 are not installed by default. Detect, install, and verify them from Settings -> CLI Environment in the TUI.

Developer Setup

Use the WSL Eip conda environment for development and verification.

source ~/miniconda3/etc/profile.d/conda.sh
conda activate Eip
python -m pip install -e ".[dev]"

TUI Notes

The current TUI supports:

• four top-level pages: execution, profiles, settings, and logs
• execution command detail forms from the execution hub
• recursive path completion with a visible candidate list
• keyboard path selection via Up, Down, Tab, Right, and Esc
• optional Browse button fallback that keeps TUI path entry working when GUI dialogs are disabled or unavailable
• profile-name persistence and TUI settings persistence under the local storage root
• log level filtering plus refresh interval state on the logs page
• left/right page switching and a scrollable page body container

Current Scope

The current implementation includes:

• Git-based package installation
• CLI entrypoint and doctor
• config bootstrap and storage migration helpers
• logging setup and structured TUI runtime events
• NRRD read/write
• slicing
• morphology refinement and hotspot map
• centroid and polar export helpers
• built-in U-Net asset resolution
• nnU-Net command adapter
• Textual TUI shell with execution, profiles, settings, logs, recursive path input, and keyboard navigation