PROJECT TITLE :
Land Classification Using Remotely Sensed Data: Going Multilabel
Getting an up-to-date high-resolution description of land cowl could be a challenging task thanks to the high value and labor-intensive process of human annotation through field studies. This work introduces a radically novel approach for achieving this goal by exploiting the proliferation of remote sensing satellite imagery, permitting for the up-to-date generation of world-scale land cowl maps. We tend to propose the applying of multilabel classification, a powerful framework in machine learning, for inferring the complicated relationships between the acquired satellite pictures and therefore the spectral profiles of different types of surface materials. Introducing a drastically different approach compared to unsupervised spectral unmixing, we tend to employ up to date ground-collected knowledge from the European Environment Agency to get the label set and multispectral images from the MODIS sensor to get the spectral options, under a supervised classification framework. To validate the deserves of our approach, we tend to present results using many state-of-the-art multilabel learning classifiers and evaluate their predictive performance with respect to the amount of annotated coaching examples, with their capability to exploit examples from neighboring regions or completely different time instances. We conjointly demonstrate the applying of our technique on hyperspectral data from the Hyperion sensor for the urban land cover estimation of New York Town. Experimental results recommend that the proposed framework will achieve excellent prediction accuracy, even from a limited number of numerous training examples, surpassing state-of-the-art spectral unmixing methods.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here