Company-Project:
Algoritmy:SK
*****FOR THIS SESSION BRING WITH YOU YOUR LAPTOP OR TABLET*****
Description:
We present the software NaturaSat devoted to the identification, classification, monitoring, and evaluation of Natura 2000 habitats by Sentinel-2 multispectral data. The NaturaSat software contains various image processing techniques based on novel mathematical models, and together with vegetation data, it makes a suitable facility for all requirements of habitat exploration. The semi-automatic and automatic segmentation methods are implemented to identify the habitat areas. The novel deep learning algorithm, natural numerical network, is implemented for habitat classification but can also be used in various research tasks or nature conservation practices, such as identifying ecosystem services and conservation value. Moreover, based on the natural numerical network, the relevancy maps are created, and it can improve many further vegetation and landscape ecology studies. The relevancy map tells us about the relevancy of the classification of the segmented area into the chosen habitat. NaturaSat provides direct access to multispectral Sentinel-2 data provided by the European Space Agency and thereby allows monitoring of Natura 2000 habitats. The monitoring process is based on calculating the statistical characteristics in the protected areas and analyzing them in time. The software was developed through intensive cooperation of botany field scientists, mathematicians, and software developers, which means that the implemented methods have a mathematical basis and are validated in field research. The NaturaSat has a user-friendly environment, for, e.g., vegetation scientists, fieldwork experts, and nature conservationists, and it is robust enough to accurately extract target unit borders, even at the habitat level. The accuracy is close to the pixel resolution; in the case of Sentinel-2 images, it is 10 m. If unmanned aerial vehicles or air-borne images are used in the software, the accuracy is rapidly pushed.