This repository provides a Dockerized setup for running the DeepWMH tool available here at https://github.com/lchdl/DeepWMH.

---

Quick Start

1. Clone the Repository

git clone <repository_url>
cd docker-deepwmh

2. Build the Docker Image

docker build -t <image_name> .

or pull the prebuilt docker image (recommended to build the image localy)

docker pull deepwmh:v1.0.1

3. Running Segmentation Tool

You can run the tool using one of the following methods:

1. Run the provided script run_deepwmh.sh to initialize the container:

   ./run_deepwmh.sh

2. or just run the following command directly in the terminal:

   docker run --rm --gpus all \
       -v /path/to/data:/data \
       -v /path/to/output:/output \
       deepwmh:v1.0.1 \ # image name
       -i /data/<flair_image> \
       -n <subject_id> \
       -m /model \
       -o /output/<subject_id> \
       -g 0

Replace /path/to/data, /path/to/output, <flair_image>, <subject_id>, and /model with the appropriate paths and values.

---

Logs and Output

• Logs are saved in the logs directory.
• Output files are saved in the output directory.

---

Notes

• Ensure that NVIDIA drivers and CUDA are installed for GPU support.
• The install_services.sh script sets up all required dependencies.
• The model file must be present in the source directory, as it is copied from the host to the container during execution.