OFFER DEADLINE01/09/2018 12:30 - Europe/Brussels
EU RESEARCH FRAMEWORK PROGRAMMEH2020 / Marie Skłodowska-Curie Actions
ORGANISATION/COMPANYInternational Project Office
DEPARTMENTPromotion & Advisory Unit
Professor, from the Department of Optics at the University of Granada, welcomes postdoctoral candidates interested in applying for a Marie Skłodowska-Curie Individual Fellowships (MSCA-IF) in this university. Applicants must comply with the Mobility Rule (more information in the participant guide: http://sl.ugr.es/097k).
The Color Imaging Lab (http://colorimaginglab.ugr.es) is a research group that belongs to the Optics Department at the University of Granada (ranked among the best 400 universities in the world and the second in Spain, according to the Shanghai ranking 2016).
In this department we have carried out research into both classical Colorimetry (e.g. color differences) and Color Vision (e.g. chromatic discrimination) since the beginning of the 1970s. In the 1990s we became interested in both human and computational color constancy. From its origin in 2000 the Color Imaging Lab has focused on the spectrum based color research and novel methods for spectral data analysis and measurement. Our current research topics are: spectral imaging, spectral estimation algorithms, high dynamic range imaging, de-weathering algorithms, human perception, color vision, visual saliency, polarimetric&spectral imaging, new spectral sensors.
In the research group we have four senior permanent staff and several Ph.D. and Post-doc students. Our laboratory is well-equipped for spectral color research with several spectral cameras and spectrometers as well as eye-trackers and a thermal camera. Our university has also the facility to use color and spectral cameras on drones.
Our group is a consortium member of the Erasmus Mundus CIMET and Erasmus+ COSI-master programmes. We are very active in international collaboration (i.e. half of our papers have international co-authorship). In research, we have project funding from the University of Granada, the Regional Government of Andalucia, the Spanish Government and from industrial partners.
Deep learning for the segmentation of singular objects in complex urban scenes
Our complex brain processing can retrieve useful information from the colour images we see, and is able to perform tasks such as segmenting the singular objects present in a scene, as well as identifying them. Nevertheless, this task is still an open problem for machine vision systems. Deep learning techniques are used to detect and identify certain objects in RGB images: for example traffic signs in self-driving vehicles. If we want our machine vision system to detect, segment or even classify more heterogeneous classes of objects (e.g. buildings, pedestrians, plants, asphalt, urban furniture, traffic signs, vehicles, and so on), the information present in RGB images is not enough.
Our objective is to improve the accuracy of segmentation and classification of objects present in complex outdoor scenes. We want to increase the amount of image data captured, not only capture RGB image information, but also multispectral (in the visible and near infrared ranges of the spectrum), high dynamic range (HDR), polarimetric and thermal information. With this big data captured pixel-wise, we can feed the machine learning and deep learning techniques with a much bigger amount of information to perform the objects segmentation and classification.
Multispectral imaging allows the study of the spectral signature of different classes of objects. Near-infrared imaging is especially helpful for discriminating different materials, and very useful for vegetation detection. Polarimetric imaging offers us information on the different objects’materials as well, especially close to highlight regions. Besides, when we are dealing with outdoor scenes, the illumination conditions are uncontrolled.
- Physics (PHY)
- Information Science and Engineering (ENG)
For a correct evaluation of your candidature, please send the documents below to Professor (firstname.lastname@example.org):
- Letter of recommendation (optional)