Hardening RGB-D Object Recognition Systems against Adversarial Patch Attacks. (arXiv:2309.07106v1 [cs.CV])

RGB-D object recognition systems improve their predictive performances by
fusing color and depth information, outperforming neural network architectures
that rely solely on colors. While RGB-D systems are expected to be more robust
to adversarial examples than RGB-only systems, they have also been proven to be
highly vulnerable. Their robustness is similar even when the adversarial
examples are generated by altering only the original images' colors. Different
works highlighted the vulnerability of RGB-D systems; however, there is a
lacking of technical …

adversarial attacks colors hardening information network neural network object patch recognition rgb robustness systems vulnerable