I am now involved with the technical committee 7 of the International Association for Pattern Recognition (IAPR) aims at promoting the use of pattern recognition methods in Earth observation.
Do not hesitate to follow us on:
Furthermore, we are sending a monthly newsletter referencing relevant conferences and events to the community. You can see the latest newsletter and subscribe.
We gave a presentation on better generic object counting in the framework of Visual Question Answering for Remote Sensing (RSVQA) at the virtual event of the 2020 ISPRS congress.
RSVQA is a new project which aims at changing the way we extract information from remote sensing images. If you missed the presentation, you can watch it here:
If you want to know more about RSVQA, you can consult the project’s page (including links to related articles) or download the conference paper.
On the 16th of September, I gave a workshop on Deep Learning for Remote Sensing together with Ahmed Samy Nassar from IRISA in France. This workshop was given at the Nordic Remote Sensing Conference (NoRSC) which took place in the Aarhus Institute of Advanced Studies, Denmark. In addition to Ahmed, this workshop was also prepared by Diego Marcos.
The first part of the tutorial was a general introduction on Deep learning, which lead to two practicals (one on classification, and the second on semantic segmentation) focused on remote sensing data. Both practicals were made using Jupyter notebooks, which the attendees can reuse to bootstrap their next project!
We would like to thank the 36 people who attended this workshop!
With Shivangi Srivastava, John Vargas and Devis Tuia, we recently published an article on land-use classification using ground based pictures (e.g. Google Street View images):
Fine-grained landuse characterization using ground-based pictures: a deep learning solution based on globally available data
Shivangi Srivastava, John Vargas, Sylvain Lobry, Devis Tuia
in International Journal of Geographical Information Science (IJGIS)
Good news: it is published in open access!
Ground-based pictures (both inside and outside) from Google Street View.
This work assigns a land-use class (e.g. religious place, hotel, post office, …) to buildings, which is an essential task towards better urban land management. In order to do this classification, we use the images from Google Street View pointing to a building as an input to a variable input siamese network (allowing to take into account the information coming from an unknown number of images). This network has been trained using ground truth labels extracted from OpenStreetMap (which are freely available).
I presented a selection of works done with my colleagues at the Laboratory of Geoinformation Science and Remote Sensing at Wageningen University and Research at the Φ-Week event organized by ESA.
I talked about our vision on how to use deep learning models to extract information from remote sensing data.
The conference was live-streamed and the replay is available below (my talk starts at 45′)!