Teaching iCub to recognize objects using deep Convolutional Neural Networks

Giulia Pasquale, Carlo Ciliberto, Francesca Odone, Lorenzo Rosasco, Lorenzo Natale
Proceedings of The 4th Workshop on Machine Learning for Interactive Systems at ICML 2015, PMLR 43:21-25, 2015.

Abstract

Providing robots with accurate and robust visual recognition capabilities in the real-world today is a challenge which prevents the use of autonomous agents for concrete applications. Indeed, the majority of tasks, as manipulation and interaction with other agents, critically depends on the ability to visually recognize the entities involved in a scene. At the same time, computer vision systems based on deep Convolutional Neural Networks (CNNs) are marking a breakthrough in fields as large-scale image classification and retrieval. In this work we investigate how latest results on deep learning can advance the visual recognition capabilities of a robotic platform (the iCub humanoid robot) in a real-world scenario. We benchmark the performance of the resulting system on a new dataset of images depicting 28 objects, named iCubWorld28, that we plan on releasing. As in the spirit of the iCubWorld dataset series, this has been collected in a framework reflecting the typical iCub’s daily visual experience. Moreover, in this release we provide four different acquisition sessions, to test incremental learning capabilities over multiple days. Our study addresses the question: how many objects can the iCub recognize today?

Cite this Paper

BibTeX
@InProceedings{pmlr-v43-pasquale15, title = {Teaching iCub to recognize objects using deep Convolutional Neural Networks}, author = {Pasquale, Giulia and Ciliberto, Carlo and Odone, Francesca and Rosasco, Lorenzo and Natale, Lorenzo}, booktitle = {Proceedings of The 4th Workshop on Machine Learning for Interactive Systems at ICML 2015}, pages = {21--25}, year = {2015}, editor = {Cuayáhuitl, Heriberto and Dethlefs, Nina and Frommberger, Lutz and Van Otterlo, Martijn and Pietquin, Olivier}, volume = {43}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {11 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v43/pasquale15.pdf}, url = {https://proceedings.mlr.press/v43/pasquale15.html}, abstract = {Providing robots with accurate and robust visual recognition capabilities in the real-world today is a challenge which prevents the use of autonomous agents for concrete applications. Indeed, the majority of tasks, as manipulation and interaction with other agents, critically depends on the ability to visually recognize the entities involved in a scene. At the same time, computer vision systems based on deep Convolutional Neural Networks (CNNs) are marking a breakthrough in fields as large-scale image classification and retrieval. In this work we investigate how latest results on deep learning can advance the visual recognition capabilities of a robotic platform (the iCub humanoid robot) in a real-world scenario. We benchmark the performance of the resulting system on a new dataset of images depicting 28 objects, named iCubWorld28, that we plan on releasing. As in the spirit of the iCubWorld dataset series, this has been collected in a framework reflecting the typical iCub’s daily visual experience. Moreover, in this release we provide four different acquisition sessions, to test incremental learning capabilities over multiple days. Our study addresses the question: how many objects can the iCub recognize today?} }
Endnote
%0 Conference Paper %T Teaching iCub to recognize objects using deep Convolutional Neural Networks %A Giulia Pasquale %A Carlo Ciliberto %A Francesca Odone %A Lorenzo Rosasco %A Lorenzo Natale %B Proceedings of The 4th Workshop on Machine Learning for Interactive Systems at ICML 2015 %C Proceedings of Machine Learning Research %D 2015 %E Heriberto Cuayáhuitl %E Nina Dethlefs %E Lutz Frommberger %E Martijn Van Otterlo %E Olivier Pietquin %F pmlr-v43-pasquale15 %I PMLR %P 21--25 %U https://proceedings.mlr.press/v43/pasquale15.html %V 43 %X Providing robots with accurate and robust visual recognition capabilities in the real-world today is a challenge which prevents the use of autonomous agents for concrete applications. Indeed, the majority of tasks, as manipulation and interaction with other agents, critically depends on the ability to visually recognize the entities involved in a scene. At the same time, computer vision systems based on deep Convolutional Neural Networks (CNNs) are marking a breakthrough in fields as large-scale image classification and retrieval. In this work we investigate how latest results on deep learning can advance the visual recognition capabilities of a robotic platform (the iCub humanoid robot) in a real-world scenario. We benchmark the performance of the resulting system on a new dataset of images depicting 28 objects, named iCubWorld28, that we plan on releasing. As in the spirit of the iCubWorld dataset series, this has been collected in a framework reflecting the typical iCub’s daily visual experience. Moreover, in this release we provide four different acquisition sessions, to test incremental learning capabilities over multiple days. Our study addresses the question: how many objects can the iCub recognize today?
RIS
TY - CPAPER TI - Teaching iCub to recognize objects using deep Convolutional Neural Networks AU - Giulia Pasquale AU - Carlo Ciliberto AU - Francesca Odone AU - Lorenzo Rosasco AU - Lorenzo Natale BT - Proceedings of The 4th Workshop on Machine Learning for Interactive Systems at ICML 2015 DA - 2015/06/18 ED - Heriberto Cuayáhuitl ED - Nina Dethlefs ED - Lutz Frommberger ED - Martijn Van Otterlo ED - Olivier Pietquin ID - pmlr-v43-pasquale15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 43 SP - 21 EP - 25 L1 - http://proceedings.mlr.press/v43/pasquale15.pdf UR - https://proceedings.mlr.press/v43/pasquale15.html AB - Providing robots with accurate and robust visual recognition capabilities in the real-world today is a challenge which prevents the use of autonomous agents for concrete applications. Indeed, the majority of tasks, as manipulation and interaction with other agents, critically depends on the ability to visually recognize the entities involved in a scene. At the same time, computer vision systems based on deep Convolutional Neural Networks (CNNs) are marking a breakthrough in fields as large-scale image classification and retrieval. In this work we investigate how latest results on deep learning can advance the visual recognition capabilities of a robotic platform (the iCub humanoid robot) in a real-world scenario. We benchmark the performance of the resulting system on a new dataset of images depicting 28 objects, named iCubWorld28, that we plan on releasing. As in the spirit of the iCubWorld dataset series, this has been collected in a framework reflecting the typical iCub’s daily visual experience. Moreover, in this release we provide four different acquisition sessions, to test incremental learning capabilities over multiple days. Our study addresses the question: how many objects can the iCub recognize today? ER -
APA
Pasquale, G., Ciliberto, C., Odone, F., Rosasco, L. & Natale, L.. (2015). Teaching iCub to recognize objects using deep Convolutional Neural Networks. Proceedings of The 4th Workshop on Machine Learning for Interactive Systems at ICML 2015, in Proceedings of Machine Learning Research 43:21-25 Available from https://proceedings.mlr.press/v43/pasquale15.html.

Related Material