Hi there!

We are pleased to announce that our Sentinel Hub community is growing rapidly and we are working hard to ensure the stability and performance of the system. In this issue you will find out more about our work in Land Cover Classification with eo-learn, and get more details about the presentation of Sentinel Hub Cloud Detector at the CEOS-WGCV ACIX II – CMIX workshop. Meet with us in person at ESA’s Φ-week in Frascati on November 12-16, and at the other coming events!


Classify the Area Around Your Hometown, or Even Your Country, with eo-learn

Mastering Satellite Image Data in an Open-Source Python Environment

Learn more about the initial procedures and the first steps in the machine learning (ML) pipeline for obtaining reliable results for land cover prediction. By using eo-learn, we lay down a strong and stable ground work of the full ML pipeline. It is done by preparing so-called patches of an area-of-interest, which contains all the needed information from Sentinel-2 band data. This includes cloud cover and "ground truth" reference masks. The process can run on any hardware from high-end scientific machines to personal laptops. A detailed and intuitive example in a Jupyter Notebook can help you get started with the code.
Stay tuned for Part 2 where we will process the data, train and validate a ML model, and, of course, use it to predict the land cover classes.
20440c99-e00d-11e7-a98f-06b2d989fe84%2F1541613229550-Land+Cover+Classification+with+eo-learn.pngA Sentinel-2 image blending into a map of predicted land cover classes.

Dive into details

Open Educational Resources

In the spirit of spreading Earth Observation (EO) knowledge, we have created an educational page, which is meant to bring the possibilities of the EO field closer to anyone who would like to learn more about it. Here you will be able to find useful links and tools addressing various EO subjects in one place. Its already ready to use and will only grow over time.

We have started with the use case devoted to wildfires. We expect to include other cases such as agriculture monitoring, flooding, deforestation, ice melting, landslides, urban growth and similar in the near future. To speed up the process, we would like to encourage you to help us raise awareness about remote sensing by simply spreading the message or contributing as described here.


A map of wildfire severity based on dNBR, showing the most damaged area in dark red, heavily damaged in orange and less damaged areas in yellow (Sentinel Playground).

Go to page

Meet us at ESA Φ-week 2018

The European Space Agency is organizing its biggest event to date, Φ-week, which will take place on November 12-16 in Frascati, Italy, and most of our Sentinel Hub team will be there. Aside from supporting Phi Week Bootcamp participants with Sentinel Hub services, we will be presenting our EO research work in several presentations.

For those participating as well, we are looking forward to meeting you. If you have missed the registration, you will still be able to follow the event online.
20440c99-e00d-11e7-a98f-06b2d989fe84%2F1541669274550-esa-phi-week.jpgCheck the official site during the event for the link to online streaming.

Join our presentations

Introducing Rate Limiting

To ensure the stability of the Sentinel Hub system and to guarantee performance for all users, we have introduced Rate Limiting. At the moment the limits are set high enough to hardly affect any of our users. In the next couple of weeks the limits will be enforced to expected levels for all newly established accounts. Users with existing accounts can expect enforcement of this policy at the beginning of January 2019.

In order to make it as frictionless as possible for you to integrate, we have set-up a test instance you can use. We have also upgraded our Python integration packages and will proceed with the rest of our integration examples in the near future.

More details    Contact us

Sentinel Hub Cloud Detector at CEOS-WGCV ACIX II - CMIX Workshop

Our research team participated at the "CEOS-WGCV ACIX II - CMIX: Atmospheric Correction Inter-comparison Exercise – Cloud Masking Inter-comparison Exercise” workshop where we presented our approach to cloud detection. In addition to the s2cloudless single scene cloud detection algorithm we are working on a multi-temporal cloud detection method that is showing even more promising results. This will be added to our eo-learn library when ready.

20440c99-e00d-11e7-a98f-06b2d989fe84%2F1540908778521-Multi-temporal+Cloud+Detection.jpg The prototype multi-temporal cloud detector successfully corrects mistakes made by the mono-temporal cloud detector.

Coming Events

To meet us in person and discuss your needs, feel free to send us an email to [email protected] and ensure yourself a meeting at the following events:
November 12-16Φ-week, Esrin, Itay
November 14GISDAY 2018, Pecs, Hungary
November 15-16The Fair of European Innovators in Cultural Heritage, Brussels, Belgium
November 21-2224th MARS Conference, Dubrovnik, Croatia
December 4EU Space Week, Marseille, France