• Source: Scopus
  • Calculated based on no. of publications stored in Pure and citations from Scopus

Research activity per year

If you made any changes in Pure these will be visible here soon.

Personal profile

Personal profile

I am Associate Professor at the University of Twente, faculty of Geo-Information Science and Earth Observation (ITC), department of Earth Observation Science (EOS), Enschede, The Netherlands. From 2011 to 2013 I was a Marie Curie research fellow with the project “MaleRS - Machine learning techniques for the analysis and classification of the last generation of remote sensing data”, supported by the European Commission and the Province of Trento. During the first two years of this project, I conducted my research activity at the Department of Empirical Inference, Max Planck Institute for Intelligent Systems, Tubingen, Germany. From June 2013 to August 2014, I was with the Remote Sensing Laboratory, Department of Information Engineering and Computer Science, University of Trento.

My main research interests are on the analysis of remote sensing data, machine learning, image classification, and pattern recognition. I am currently working on the analysis of images acquired by Unmanned Airborne Vehicles (UAVs) for monitoring urban areas.

Check my Google Scholar citations here.


Honours and awards

2012-Best PhD thesis on Pattern Recognition published between 2010 and 2012 awarded by GIRPR, i.e., the Italian branch of the International Association for Pattern Recognition (IAPR).

Awarded a three-year Marie Curie research fellowship for the project MaleRS.


Personal profile

Activities in research

My research activity is mainly about the development of Machine Learning (ML) techniques for the analysis and classification of the last generation of Remote Sensing (RS) images. The recent advances in optical sensors and satellite technology resulted in the development of systems capable to acquire remote-sensing images with very high resolution both in the geometrical and spectral domain. Satellite multispectral scanners can acquire images with Very High geometrical Resolution (VHR), which is in the order or smaller than one meter. These images can be used for precisely characterizing the type and the geometrical properties of the objects on the ground (e.g., buildings, streets, agriculture fields). Hyperspectral sensors can acquire images characterized by hundreds of bands associated to narrow spectral channels. These data allow one to precisely measure the spectral signature of the different materials on the ground, which can be used for discriminating similar objects in the scene under investigation and for estimating biophysical parameters (e.g., forest biomass).

VHR and hyperspectral images provide very useful information for several applications related to the monitoring of natural resources and human structures. However, in order to develop real-world applications, it is necessary to define automatic techniques that are able to effectively and efficiently analyse these high dimensional data. My research activity is focused on RS image classification, which is at the basis of most of the applications related to the monitoring of the environment. Image classification is devoted to automatically recognize the different land-cover types present on the ground and to produce a thematic map of the investigated area. However, the huge amount of information associated with VHR and hyperspectral RS images makes the classification problem very complex and the available techniques are still inadequate to properly analyse these kinds of data. For this reason, my general research objective is to develop novel ML techniques for the analysis and the classification of VHR and hyperspectral images, in order to improve the capability to automatically extract useful information captured from these data and to exploit it in real applications.

In the past years, I focused on Support Vector Machines (SVM) and kernel methods for addressing classification problems under different operative conditions (e.g., in the presence of small, biased or even noisy training sets). I investigated the use of different learning paradigms, i.e., supervised, semi supervised and active learning for the classification of RS images.

I am currently investigating deep learning techniques for the analysis of very high resolution images in the context of different applications.

Here I provide some keywords for describing the main aspects of machine learning and remote sensing I am interested in:

  • UAV images and photogrammetric point clouds;
  • Multispectral VHR satellite images;
  • Hyperspectral images;
  • deep learning;
  • convolutional neural networks;
  • SVM and kernel methods;
  • Supervised, semi supervised and active learning;
  • Transfer learning and domain adaptation;
  • Feature selection/extraction;
  • Classification map accuracy assessment;
  • Mapping of agriculture fields;
  • Forest inventories;
  • Ground sample collection optimization.


Dive into the research topics where Claudio Persello is active. These topic labels come from the works of this person. Together they form a unique fingerprint.
  • 1 Similar Profiles


Recent external collaboration on country/territory level. Dive into details by clicking on the dots or