132128 20240731103339.0 doi 10.1109/ICRA48891.2023.10160606 sideral 137285 ART-2023-137285 eng Mur-Labadia, Lorenzo Universidad de Zaragoza Bayesian deep learning for affordance segmentation in images 2023 Access copy available to the general public Unrestricted Affordances are a fundamental concept in robotics since they relate available actions for an agent depending on its sensory-motor capabilities and the environment. We present a novel Bayesian deep network to detect affordances in images, at the same time that we quantify the distribution of the aleatoric and epistemic variance at the spatial level. We adapt the Mask-RCNN architecture to learn a probabilistic representation using Monte Carlo dropout. Our results outperform the state-of-the-art of deterministic networks. We attribute this improvement to a better probabilistic feature space representation on the encoder and the Bayesian variability induced at the mask generation, which adapts better to the object contours. We also introduce the new Probability-based Mask Quality measure that reveals the semantic and spatial differences on a probabilistic instance segmentation model. We modify the existing Probabilistic Detection Quality metric by comparing the binary masks rather than the predicted bounding boxes, achieving a finer-grained evaluation of the probabilistic segmentation. We find aleatoric variance in the contours of the objects due to the camera noise, while epistemic variance appears in visual challenging pixels. info:eu-repo/grantAgreement/ES/MICINN-AEI/PID2021-125209OB-I00 info:eu-repo/grantAgreement/EUR/MICINN/TED2021-129410B-I00 info:eu-repo/semantics/openAccess All rights reserved http://www.europeana.eu/rights/rr-f/ 1.62 2023 Artificial Intelligence 2023 Software 2023 Electrical and Electronic Engineering 2023 Control and Systems Engineering 2023 6.8 2023 info:eu-repo/semantics/conferenceObject info:eu-repo/semantics/acceptedVersion Martinez-Cantin, Rubén Universidad de Zaragoza (orcid)0000-0002-6741-844X Guerrero, José J. Universidad de Zaragoza (orcid)0000-0001-5209-2267 5007 520 Universidad de Zaragoza Dpto. Informát.Ingenie.Sistms. Área Ingen.Sistemas y Automát. 2023 (2023), 6981-6987 Proc. - IEEE Int. Conf. Robot. Autom. Proceedings - IEEE International Conference on Robotics and Automation 1050-4729 1425642 http://zaguan.unizar.es/record/132128/files/texto_completo.pdf Postprint info:eu-repo/date/embargoEnd/2024-07-04 3128617 http://zaguan.unizar.es/record/132128/files/texto_completo.jpg?subformat=icon icon Postprint info:eu-repo/date/embargoEnd/2024-07-04 oai:zaguan.unizar.es:132128 articulos driver 2024-07-31-09:49:35 ARTICLE