Tracasa succeeds in identifying roads and highways using images from Sentinel satellites, super-resolved with Artificial Intelligence

The Government of Navarra’s public company presents its SEN2ROAD model, which paves the way for applications like the creation of quick maps of specific sites and logistical and mobility planning

Image of Bilbao featuring road segmentation.

Tracasa, a public company of the Government of Navarra, has managed to identify roads and highways of various cities on the Iberian Peninsula using images from the Sentinel 2A satellites, super-resolved employing Artificial Intelligence techniques. Tracasa pulled this feat off with its SEN2ROAD model, which opens the door to many usages in the future, from creating quick maps of specific sites to logistical and mobility planning. SEN2ROAD is now available for viewing on the Tracasa website, in the Innovative stories section, via the following link.

“It was a challenge to obtain an extraction of roads and highways using images from the European Space Agency’s Sentinel 2A satellites. These freely accessible images usually feature a spatial resolution of 10 meters per pixel, which, in many cases is insufficient for a correct segmentation of the terrain, so we had to rely on our SENX4 super-resolution model to enhance the definition and obtain more precise segmentation”, explained Tracasa’s R&D head Carlos Aranda.

Within the SENX4 model, Tracasa uses Deep Learning techniques in the photogrammetry environment. These techniques use DL algorithms to mimic and improve the human ability to process and recognize patterns in images, but without performing post-processing actions to enhance their resolution.

“In cartography, as a general rule, the quality and even existence of layers of roads and highways declines as the size of the city decreases, or if it is located in a developing country”, explained Aranda, who stressed that, in this regard, the SEN2ROAD model, based on free-access Sentinel images, can offer layers of urban roads and highways of any location, with a significant reduction in costs, and affording greater access to these services. “For example, governments and companies can use this new model in actions for rapid road mapping, logistical and mobility planning, disaster response management of all kinds, and the analysis of urban development and carbon footprints”.

Vectorization for multiple uses

The SEN2ROAD model constitutes a valuable asset in terms of spatial representation, but Tracasa wanted to go a step further, making this information applicable to decision-making in, for example, the area of communication planning.

“With a view to making the most of the segmented information through SEN2ROAD, we went forward with an optimized raster vectorization project. In this way, through our SEG2VECTOR algorithm library, we have converted the information from the pixels that represent the roads into vectors, creating a network of linear segments and nodes. This information on the roads allows for the simulation of transport networks and subsequent studies based on mobility criteria”, Aranda concluded.

Innovative stories: one step ahead

In its Innovative stories Tracasa presents to society, in a series of instalments, its most cutting-edge research, those tools delivering disruptive technology and that may see major development in the coming years. “The drive to discover new paths of technological development is in Tracasa’s DNA. In this regard, our advances in terms of innovation converge at a meeting point: the Innovative stories”, Aranda specified.

After presenting SENX4, a model with which it is possible to multiply four-fold the resolution of the images from Sentinel 2A satellites through Deep Learning techniques, Tracasa plans to soon roll out innovative new approaches to areas such as Artificial Intelligence, Data Science, automated remote sensing, geoscience analytical models, specialized digital models of the territory, 3D cities, smart city platforms, digital twins (Internet of Things), and smart mapping in the field of mobility.

A firm commitment to Artificial Intelligence

In recent years Tracasa has made a firm commitment to the use of Artificial Intelligence. In this area, the public company of the Government of Navarra has carried out, in collaboration with Tracasa Instrumental, 16 projects related to Artificial Intelligence since 2017, 11 of them in the last year. Since 2019 Tracasa Instrumental has form part of the AI Technology Capabilities Map in Spain, and has recently incorporated its services into the Digital Innovation Pole promoted by the Government of Navarra.