Version 1.2 Release Notes

The latest version of EOTDL is now available 🥳. And with it we bring a set of improvements, bug fixes and changes in order to improve your experience.

Make sure to update your library/CLI to the latest version to take advantage of all the new features and improvements. pip install eotdl --upgrade

New Features

New metadata format for Q0 datasets/models

The main new feature that we are introducing in v1.2 is how you define the metadata for your Q0 datasets/models. Until now, a metadata.yml file with information about the authors, license, etc. was required. Now, you can define this information in a README.md file, which can also contain the description of the dataset/model that is shown in the UI.

Before:

name: dataset-nmae
license: dataset-license
source: https://example-slurce.com
authors:
  - author 1
  - author 2

Now:

---
name: dataset-nmae
license: dataset-license
thumbnail: https://example-thumbnail.com
source: https://example-slurce.com
authors:
  - author 1
  - author 2
---

# Title for your page

some content ...

As you can see, the same metadata than before is now defined in the README.md file as frontmatter. Additionally, a new metadata field called thumbnail is optional to show a preview of the dataset/model in the UI.

This file is supposed to be in sync with the metadata shown in the UI, which can be modified by the user. Two new flags have been added to the ingest command to either force the update of the metadata in the README.md file or to sync the metadata in the README.md file with the one in the UI.

eotdl dataset ingest -p path-to-dataset -f # force update the metadata in the README.md file
eotdl dataset ingest -p path-to-dataset -s # sync the metadata in the README.md file with the one in the UI

This works as well with models.

Upgrade button in UI

The datasets page now show an Upgrade button that will redirect you to an example notebook on your cloud workspace with an example on how to upgrade datasets fro Q0 to Q1, and from Q1 to Q2.

New Third Party Applications page

The UI now has a new page that shows a list of third party applications that are integrated with EOTDL. This page is accessible here.

Get in touch with us if you want your application to be integrated with EOTDL and listed here for users to discover and use.

API Key access

A new method to interact with the EOTDL via API Key has been implemented, so you don't have to manually login to the library/CLI. This is useful for CI/CD pipelines and other automated workflows.

For the moment, we are not making it available to every user to request an API Key. However, if you are interested in using this feature, please get in touch with us and will gladly provide you with an API Key.

New datasets and Models

And, of course, new datasets and model have been included.

Changes

• The metadata.yml file is no longer required. You can define the metadata in the README.md file.
• The README.md file will not be ingested along with your dataset/model, and it will be auto-generated when you download a dataset.
• Increased login time in the library/CLI.
• The navigation bar in the UI has been improved with icons and direct links to the cloud workspace and third party applications page.

Bug Fixes

• Sync issues with local and remote Q0 metadata are now solved due to the new metadata ingestion mechanism.

Miscellaneous

EOTDL is now the official datasets repository for the AI4EO Challenges Platform. Starting with the MapYourCity challenge, all the datasets for future challenges will be hosted in EOTDL.

We encourage participants to share their models as well in EOTDL, so that the community can benefit from the work done in the challenges.