Training Lightweight CNNs for Human-Nanodrone Proximity Interaction from Small Datasets using Background Randomization
Marco Ferri
Dario Mantegazza
Elia Cereda
Nicky Zimmerman
L. Gambardella
Daniele Palossi
Jérôme Guzzi
Alessandro Giusti

Abstract
We consider the task of visually estimating the pose of a human from images acquired by a nearby nano-drone; in this context, we propose a data augmentation approach based on synthetic background substitution to learn a lightweight CNN model from a small real-world training set. Experimental results on data from two different labs proves that the approach improves generalization to unseen environments.
View on arXivComments on this paper