A commercial robot, trained by its manufacturer to recognize a predefined number and type of objects, might be used in many settings, that will in general differ in their illumination conditions, background, type and degree of clutter, and so on. Recent computer vision works tackle this generalization issue through domain adaptation methods, assuming as source the visual domain where the system is trained and as target the domain of deployment. All approaches assume to have access to images from all classes of the target during training, an unrealistic condition in robotics applications. We address this issue proposing an algorithm that takes into account the specific needs of robot vision. Our intuition is that the nature of the domain shift experienced mostly in robotics is local. We exploit this through the learning of maps that spatially ground the domain and quantify the degree of shift, embedded into an end-to-end deep domain adaptation architecture. By explicitly localizing the roots of the domain shift we significantly reduce the number of parameters of the architecture to tune, we gain the flexibility necessary to deal with subset of categories in the target domain at training time, and we provide a clear feedback on the rationale behind any classification decision, which can be exploited in human-robot interactions. Experiments on two different settings of the iCub World database confirm the suitability of our method for robot vision.

Adaptive Deep Learning through Visual Domain Localization / Angeletti, Gabriele; Caputo, Barbara; Tommasi, Tatiana. - STAMPA. - (2018), pp. 7135-7142. (Intervento presentato al convegno IEEE International Conference on Robotics and Automation (ICRA) tenutosi a Brisbane; Australia nel May 2018) [10.1109/ICRA.2018.8460650].

Adaptive Deep Learning through Visual Domain Localization

Gabriele Angeletti;Barbara Caputo;Tatiana Tommasi
2018

Abstract

A commercial robot, trained by its manufacturer to recognize a predefined number and type of objects, might be used in many settings, that will in general differ in their illumination conditions, background, type and degree of clutter, and so on. Recent computer vision works tackle this generalization issue through domain adaptation methods, assuming as source the visual domain where the system is trained and as target the domain of deployment. All approaches assume to have access to images from all classes of the target during training, an unrealistic condition in robotics applications. We address this issue proposing an algorithm that takes into account the specific needs of robot vision. Our intuition is that the nature of the domain shift experienced mostly in robotics is local. We exploit this through the learning of maps that spatially ground the domain and quantify the degree of shift, embedded into an end-to-end deep domain adaptation architecture. By explicitly localizing the roots of the domain shift we significantly reduce the number of parameters of the architecture to tune, we gain the flexibility necessary to deal with subset of categories in the target domain at training time, and we provide a clear feedback on the rationale behind any classification decision, which can be exploited in human-robot interactions. Experiments on two different settings of the iCub World database confirm the suitability of our method for robot vision.
2018
IEEE International Conference on Robotics and Automation (ICRA)
robotics; computer vision; machine learning
04 Pubblicazione in atti di convegno::04b Atto di convegno in volume
Adaptive Deep Learning through Visual Domain Localization / Angeletti, Gabriele; Caputo, Barbara; Tommasi, Tatiana. - STAMPA. - (2018), pp. 7135-7142. (Intervento presentato al convegno IEEE International Conference on Robotics and Automation (ICRA) tenutosi a Brisbane; Australia nel May 2018) [10.1109/ICRA.2018.8460650].
File allegati a questo prodotto
File Dimensione Formato  
Angeletti_Postprint-Adaptive_2018.pdf

accesso aperto

Note: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8460650
Tipologia: Documento in Post-print (versione successiva alla peer review e accettata per la pubblicazione)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 2.06 MB
Formato Adobe PDF
2.06 MB Adobe PDF
Angeletti_Adaptive_2018.pdf

solo gestori archivio

Tipologia: Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 1.58 MB
Formato Adobe PDF
1.58 MB Adobe PDF   Contatta l'autore

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1092189
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 14
  • ???jsp.display-item.citation.isi??? 9
social impact