Fully Convolutional Siamese Networks for Change Detection

Rodrigo Caye Daudt (1, 2) , Bertrand Le Saux (1) , Alexandre Boulch (1)

1 DTIS, ONERA, Université Paris Saclay (COmUE) [Palaiseau]
2 LTCI - Laboratoire Traitement et Communication de l'Information

This paper presents three fully convolutional neural network architectures which perform change detection using a pair of coregistered images. Most notably, we propose two Siamese extensions of fully convolutional networks which use heuris-tics about the current problem to achieve the best results in our tests on two open change detection datasets, using both RGB and multispectral images. We show that our system is able to learn from scratch using annotated change detection images. Our architectures achieve better performance than previously proposed methods, while being at least 500 times faster than related systems. This work is a step towards efficient processing of data from large scale Earth observation systems such as Copernicus or Landsat.

Mots clés: Index Terms— Change detection; supervised machine learning; fully convolutional networks; Earth observation

Domaines: Traitement des images [eess.IV]; Réseau de neurones [cs.NE]; Vision par ordinateur et reconnaissance de formes [cs.CV]

Rodrigo Caye Daudt, Bertrand Le Saux, Alexandre Boulch. Fully Convolutional Siamese Networks for Change Detection. IEEE International Conference on Image Processing, Oct 2018, Athens, Greece. ⟨hal-01824557⟩