This site should be considered in BETA status and is an active WIP!
The current design makes it easy to quickly provide feedback and highlight issues by utilizing the hypothes.is library.
On the right-hand side, you can access a bar to see comments for the current page and create your annotations by highlighting text and clicking on the
If you have annotated something on the website, please make sure to set it to public so that we can read it as well 😉.
As we aren’t notified about annotations to the website, please inform us via GitHub issues or via e-mail that you added suggestions.
See Contributing for more ways to contribute!
The main goal of this documentation is to introduce the multi-modal BigEarthNet dataset and to make it more accessible to others by providing an interactive dataset website. By providing visual and interactive examples, users can see how the data is structured and how it can be used in different contexts. Also, the website contains background information and explains how to avoid common pitfalls to lower the probability of different users repeating similar mistakes multiple times. To aggregate the knowledge of the community, a core feature of the dataset website is the ability to quickly provide feedback and contribute to the living document in an open and centralized manner. Doing so, will (hopefully) lead to more robust and reliable research down the line.
Specifically, by providing the following reference, our goals are to:
Provide interested users a high-level overview of the BigEarthNet dataset
Where does the data come from?
What are patches or tiles?
How is the dataset structured?
How do I work with the metadata?
How can I visualize the data?
How do I work with different spatial resolutions?
Introduce libraries that make it easier to work with BigEarthNet?
Give behind the scenes details about design decisions that are necessary to work with the data
Provide different optimized approaches to access the dataset to reduce compute-time and power usage greatly
Heavily depends on the workload but can lead to significant speed-ups (5–10x)