This project contains material to support the sessions carried out among the DeepHealth partners to complete different tasks and reach different goals within the DeepHealth project.
We plan to have several sessions in order to show other partners in the project how to install, configure and use the libraries developed in the DeepHealth project. Further support in addition to these sessions will be provided in order to reach all platform providers integrate the libraries into their application platforms.
For this first session the following material is available in order to help DeepHealth partners and any other interested people in preparing the complete pipeline. What you can see here has been prepared starting from the use case pipeline prepared by UNIMORE team. But the purpose of this alternative is to have an installation of the EDDL in a conda environment named eddl ready to be used in any other project or task.
-
Shell-scripts used to do the installation of the libraries and the pipeline example. Must be executed in the order they appear:
- prep-installation-1.sh
- prep-installation-2.sh
- prep-installation-3.sh
- install-eddl-from-source.sh
- install-opencv-from-source.sh
- install-ecvl-from-source.sh
- build_pipeline_2.sh
- env-vars.sh This file contains definition of environment variables used in the in the previous shell-scripts.
-
Video tutorial 1 where Anaconda3, EDDL, OpenCV, ECVL and use case pipeline are installed on a Linux container with Ubuntu 18.04:
- Video 1a shows how prep-installation-1.sh runs.
- Video 1b shows how prep-installation-2.sh runs.
- Video 1c shows how prep-installation-3.sh runs.
- Video 1d shows how install-eddl-from-source.sh runs.
- Video 1e shows how install-opencv-from-source.sh runs.
- Video 1f shows how install-ecvl-from-source.sh runs.
- Video 1g shows how build_pipeline_2.sh runs.
-
Video tutorial 2 where code changes are shown in order to use the YAML files Video 2
-
Dataset extracted from PADCHEST split in two ZIP files:
-
Alternative installation procedure for non-privileged users when conda is already available in the Linux system (no videos were prepared):
- prep-installation.sh
- install-eddl-from-source.sh
- install-opencv-from-source.sh
- install-ecvl-from-source.sh
- build_pipeline_2.sh
- env-vars.sh This file contains definition of environment variables used in the in the previous shell-scripts.
The main difference is in the first step, now only one shell-script. There are small differences in the other files, the component calib3d has been added to OpenCV, this solved the problem with the method cv::initDistortRectifyMap() when compiling the ECVL.
IMPORTANT: if a conda installation is not available in the system and the base conda environment is not activated before starting the process, then the execution of the shell-scripts will fail.
-
Presentations:
- UC12 & Pneumothorax in C++ (UNIMORE)
- Pneumothorax: demo for downloading the dataset, download the masks (ground truth), cloning the pipeline repository, compile and execute the setup program, which copies the masks into the right folders, compile and execute the training program. Slides
-
Padchest in C++ (UPV) Slides
-
UC12 & Pneumothorax in Python (UNIMORE)
-
Padchest in Python (UPV)
-
Configuration issues and programming:
- Testing pipelines run on partners’ computers and giving the necessary support
-
Presentation of results
-
Presentation and demo of the back-end
-
Presentation and demo of the front-end
-
Configuration and programming issues (if required by any partner)
- T.B.A. it will be HPC+Cloud oriented