Learning Depth-Aware Deep Representations for Robotic Perception
Lorenzo Porzi, Samuel Rota Bulò, Adrián Peñate Sánchez, Elisa Ricci, Francesc Moreno-Noguer
Abstract
Exploiting RGB-D data by means of convolutional neural networks (CNNs) is at the core of a number of robotics applications, including object detection, scene semantic segmentation, and grasping. Most existing approaches, however, exploit RGB-D data by simply considering depth as an additional input channel for the network. In this paper we show that the performance of deep architectures can be boosted by introducing DaConv, a novel, general-purpose CNN block which exploits depth to learn scale-aware feature representations. We demonstrate the benefits of DaConv on a variety of robotics oriented tasks, involving affordance detection, object coordinate regression, and contour detection in RGB-D images. In each of these experiments we show the potential of the proposed block and how it can be readily integrated into existing CNN architectures.
BibTeX
@inproceedings{ral2017_learningdepthawa,
title = {Learning Depth-Aware Deep Representations for Robotic Perception},
author = {Lorenzo Porzi and Samuel Rota Bulò and Adrián Peñate Sánchez and Elisa Ricci and Francesc Moreno-Noguer},
booktitle = {RA-L 2017},
year = {2017}
}