Depth Value Pre-Processing for Accurate Transfer Learning Based RGB-D Object Recognition

Publikation: Forskning - peer reviewKonferenceartikel i proceeding

Abstrakt

Object recognition is one of the important tasks in computer vision which has found enormous applications.Depth modality is proven to provide supplementary information to the common RGB modality for objectrecognition. In this paper, we propose methods to improve the recognition performance of an existing deeplearning based RGB-D object recognition model, namely the FusionNet proposed by Eitel et al. First, we showthat encoding the depth values as colorized surface normals is beneficial, when the model is initialized withweights learned from training on ImageNet data. Additionally, we show that the RGB stream of the FusionNetmodel can benefit from using deeper network architectures, namely the 16-layered VGGNet, in exchange forthe 8-layered CaffeNet. In combination, these changes improves the recognition performance with 2.2% incomparison to the original FusionNet, when evaluating on the Washington RGB-D Object Dataset.
Luk

Detaljer

Object recognition is one of the important tasks in computer vision which has found enormous applications.Depth modality is proven to provide supplementary information to the common RGB modality for objectrecognition. In this paper, we propose methods to improve the recognition performance of an existing deeplearning based RGB-D object recognition model, namely the FusionNet proposed by Eitel et al. First, we showthat encoding the depth values as colorized surface normals is beneficial, when the model is initialized withweights learned from training on ImageNet data. Additionally, we show that the RGB stream of the FusionNetmodel can benefit from using deeper network architectures, namely the 16-layered VGGNet, in exchange forthe 8-layered CaffeNet. In combination, these changes improves the recognition performance with 2.2% incomparison to the original FusionNet, when evaluating on the Washington RGB-D Object Dataset.
OriginalsprogEngelsk
TitelInternational Joint Conference on Computational Intelligence
Publikationsdato8 aug. 2017
StatusAccepteret/In press - 8 aug. 2017
PublikationsartForskning
Peer reviewJa
BegivenhedInternational Joint Conference on Computational Intelligence - Funchal, Portugal
Varighed: 1 nov. 20173 nov. 2017
Konferencens nummer: 9
http://www.ijcci.org/

Konference

KonferenceInternational Joint Conference on Computational Intelligence
Nummer9
LandPortugal
ByFunchal
Periode01/11/201703/11/2017
Internetadresse

Kort

Download-statistik

Ingen data tilgængelig
ID: 261392233