Building detection in very high resolution multispectral data with deep learning features

Maria Vakalopoulou (1, 2) , Konstantinos Karantzalos (2) , Nikos Komodakis (3, 1) , Nikos Paragios (4)

1 ENPC - École des Ponts ParisTech
2 Remote Sensing Laboratory
3 imagine [Marne-la-Vallée]
4 CVN - Centre de vision numérique

The automated man-made object detection and building extraction from single satellite images is, still, one of the most challenging tasks for various urban planning and monitoring engineering applications. To this end, in this paper we propose an automated building detection framework from very high resolution remote sensing data based on deep convolu-tional neural networks. The core of the developed method is based on a supervised classification procedure employing a very large training dataset. An MRF model is then responsible for obtaining the optimal labels regarding the detection of scene buildings. The experimental results and the performed quantitative validation indicate the quite promising potentials of the developed approach.

Mots clés: AlexNet; deep convolutional networks; Remote Sensing; man made objects extraction

Domaines: Interfaces continentales, environnement; Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Maria Vakalopoulou, Konstantinos Karantzalos, Nikos Komodakis, Nikos Paragios. Building detection in very high resolution multispectral data with deep learning features. Geoscience and Remote Sensing Symposium (IGARSS), 2015 IEEE International, Jul 2015, Milan, Italy. pp.1873-1876, ⟨10.1109/IGARSS.2015.7326158⟩. ⟨hal-01264084⟩