Filters








469 Hits in 7.0 sec

Learning to Disambiguate Strongly Interacting Hands via Probabilistic Per-pixel Part Segmentation [article]

Zicong Fan, Adrian Spurr, Muhammed Kocabas, Siyu Tang, Michael J. Black, Otmar Hilliges
2021 arXiv   pre-print
The method consists of two interwoven branches that process the input imagery into a per-pixel semantic part segmentation mask and a visual feature volume.  ...  To do so, the part probabilities are merged with the visual features and processed via fully-convolutional layers.  ...  While MJB is a part-time employee of Amazon, his research was performed solely at, and funded solely by, Max Planck. MJB has financial interests in Amazon, Datagen Technologies, and Meshcapade GmbH.  ... 
arXiv:2107.00434v2 fatcat:frrzchccbja4pclwkh4ebhposi

Learning To Disambiguate Strongly Interacting Hands via Probabilistic Per-Pixel Part Segmentation

Zicong Fan, Adrian Spurr, Muhammed Kocabas, Siyu Tang, Michael J. Black, Otmar Hilliges
2021 2021 International Conference on 3D Vision (3DV)  
When estimating the 3D pose of interacting hands, state-of-the-art methods struggle to disambiguate the appearance of the two hands and their parts.  ...  Our model, DIGIT, reduces the ambiguity by predicting and leveraging a probabilistic part segmentation volume (2.1) to produce reliable pose estimates even when the two hands are in direct contact and  ...  While MJB is a part-time employee of Amazon, his research was performed solely at, and funded solely by, Max Planck. MJB has financial interests in Amazon, Datagen Technologies, and Meshcapade GmbH.  ... 
doi:10.1109/3dv53792.2021.00011 fatcat:l6ozyteqv5dixa7x6vfgnponpq

Context based object categorization: A critical survey

Carolina Galleguillos, Serge Belongie
2010 Computer Vision and Image Understanding  
Context information, based on the interaction among objects in the scene or global scene statistics, can help successfully disambiguate appearance inputs in recognition tasks.  ...  The goal of object categorization is to locate and identify instances of an object category within an image.  ...  Also, other machine learning models will be considered for a better integration of context features.  ... 
doi:10.1016/j.cviu.2010.02.004 fatcat:3ee2st4tffbnplewyrq5o4i5rm

Multi-class object localization by combining local contextual interactions

Carolina Galleguillos, Brian McFee, Serge Belongie, Gert Lanckriet
2010 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition  
Furthermore, we study context as (i) part of recognizing objects in images and (ii) as an advocate for label agreement to disambiguate object identity in recognition systems.  ...  We analyze the contributions and trade -offs of integrating context and investigate contextual interactions between pixels, regions and objects in the scene.  ...  Given an image I, its corresponding segments S 1 , . . . , S k , and probabilistic per-segment labels p(c i |S i ) (as in [65] ), we wish to find segment labels c 1 , . . . , c k ∈ C such that all agree  ... 
doi:10.1109/cvpr.2010.5540223 dblp:conf/cvpr/GalleguillosMBL10 fatcat:asm6h3hix5fufkipt56voasqcu

Learning to Regress Bodies from Images using Differentiable Semantic Rendering [article]

Sai Kumar Dwivedi, Nikos Athanasiou, Muhammed Kocabas, Michael J. Black
2022 arXiv   pre-print
Learning to regress 3D human body shape and pose (e.g.  ...  ~SMPL parameters) from monocular images typically exploits losses on 2D keypoints, silhouettes, and/or part-segmentation when 3D training data is not available.  ...  On the other hand, relying too strongly on 3D priors introduces bias. To circumvent this problem, recent approaches [30, 34, 50] propose to use part-segmentations or silhouettes.  ... 
arXiv:2110.03480v2 fatcat:6oi25y5pw5bgdbhzunz47rh2ny

Processing reduced word-forms in speech perception using probabilistic knowledge about speech production

Holger Mitterer, James M. McQueen
2009 Journal of Experimental Psychology: Human Perception and Performance  
Prior to disambiguation, listeners' fixations were drawn to /t/-final words more when boven than when naast followed the ambiguous sequences.  ...  We thus argue that probabilistic knowledge about the effect of following context in speech production is used prelexically in perception to help resolve lexical ambiguities caused by continuous-speech  ...  On the one hand, schwa deletion in a word such as president is likely to be independent of the word's segmental context.  ... 
doi:10.1037/a0012730 pmid:19170486 fatcat:crzskrhbqfhalpmbjerrhscvnm

Using Motion and Internal Supervision in Object Recognition [article]

Daniel Harari
2018 arXiv   pre-print
In the second and main part of the study we develop methods for using specific types of visual motion to solve two difficult problems in unsupervised visual learning: learning to recognize hands by their  ...  We use our conclusions in this part to propose a model for several aspects of learning by human infants from their visual environment.  ...  Thus, we can strongly believe that infants use motion as a major source of information for learning hands.  ... 
arXiv:1812.05455v1 fatcat:peih4lfnnjacncpvgeutjbk7ue

Pixel Objectness: Learning to Segment Generic Objects Automatically in Images and Videos [article]

Bo Xiong, Suyog Dutt Jain, Kristen Grauman
2018 arXiv   pre-print
We propose an end-to-end learning framework for segmenting generic objects in both images and videos.  ...  Thus we propose ways to exploit weakly labeled data for learning dense foreground segmentation.  ...  ACKNOWLEDGEMENTS This research is supported in part by ONR YIP N00014-12-1-0754, an AWS Machine Learning Research Award, and DARPA Lifelong Learning Machines.  ... 
arXiv:1808.04702v2 fatcat:jvin6gvjwndehjcbz3n336e5f4

Pixel Objectness: Learning to Segment Generic Objects Automatically in Images and Videos

2018 IEEE Transactions on Pattern Analysis and Machine Intelligence  
We propose an end-to-end learning framework for segmenting generic objects in both images and videos.  ...  Thus we propose ways to exploit weakly labeled data for learning dense foreground segmentation.  ...  ACKNOWLEDGEMENTS This research is supported in part by ONR YIP N00014-12-1-0754, an AWS Machine Learning Research Award, and the DARPA Lifelong Learning Machines project.  ... 
doi:10.1109/tpami.2018.2865794 pmid:30130176 fatcat:nmx3kdvw7vcslfgb623lwc4ose

Fundamental Strategies for Solving Low-Level Vision Problems

Marshall F. Tappen
2011 IPSJ Transactions on Computer Vision and Applications  
Solutions to low-level problems can be broadly group according to how they propagate local information to global representations.  ...  Understanding these categorizations is useful because they offer guidance on how tools like machine learning can be implemented in these systems.  ...  that the model is limited to the pairwise interactions.  ... 
doi:10.2197/ipsjtcva.3.95 fatcat:m6x6xitzhnbztmqhrskssdbf44

People Watching: Human Actions as a Cue for Single View Geometry

David F. Fouhey, Vincent Delaitre, Abhinav Gupta, Alexei A. Efros, Ivan Laptev, Josef Sivic
2014 International Journal of Computer Vision  
These constraints are then used to improve single-view 3D scene understanding approaches.  ...  Our method builds upon recent advances in still-image pose estimation to extract functional and geometric constraints on the scene.  ...  Acknowledgments: This work was supported by NSF Graduate Research and NDSEG Fellowships to DF, and by ONR-MURI N000141010934, NSF IIS-1320083, the MSR-INRIA laboratory, the EIT-ICT labs, Google, ERC Activia  ... 
doi:10.1007/s11263-014-0710-z fatcat:auj4lihccvesnnl32wlq6zd34q

People Watching: Human Actions as a Cue for Single View Geometry [chapter]

David F. Fouhey, Vincent Delaitre, Abhinav Gupta, Alexei A. Efros, Ivan Laptev, Josef Sivic
2012 Lecture Notes in Computer Science  
These constraints are then used to improve single-view 3D scene understanding approaches.  ...  Our method builds upon recent advances in still-image pose estimation to extract functional and geometric constraints on the scene.  ...  Acknowledgments: This work was supported by NSF Graduate Research and NDSEG Fellowships to DF, and by ONR-MURI N000141010934, NSF IIS-1320083, the MSR-INRIA laboratory, the EIT-ICT labs, Google, ERC Activia  ... 
doi:10.1007/978-3-642-33715-4_53 fatcat:ma4fdudovjf27fxvz7eo2alwhu

Context-Based Bayesian Intent Recognition

R. Kelley, A. Tavakkoli, C. King, A. Ambardekar, M. Nicolescu, M. Nicolescu
2012 IEEE Transactions on Autonomous Mental Development  
One of the foundations of social interaction among humans is the ability to correctly identify interactions and infer the intentions of others.  ...  To build robots that reliably function in the human social world, we must develop models that robots can use to mimic the intent recognition skills found in humans.  ...  Low-Level Recognition via Hidden Markov Models As mentioned above, our system uses HMMs to model activities that consist of a number of parts that have intentional significance.  ... 
doi:10.1109/tamd.2012.2211871 fatcat:kdw6kdirhbfxfk2wvq4ne7bgoe

Interleaved Object Categorization and Segmentation

B. Leibe, B. Schiele
2003 Procedings of the British Machine Vision Conference 2003  
These probabilities correspond to the per-pixel confidence the system has in its recognition and segmentation result.  ...  In addition, it returns a per-pixel confidence estimate, specifying how much this segmentation can be trusted.  ... 
doi:10.5244/c.17.78 dblp:conf/bmvc/LeibeS03 fatcat:tdpwrxy44be73ctuxzdeyop26q

Unsupervised construction of human body models

Thomas Walther, Rolf P. Würtz
2018 Cognitive Systems Research  
Unsupervised learning of a generalizable model of the visual appearance of humans from video data is of major importance for computing systems interacting naturally with their users and others.  ...  The models from many videos are integrated into a meta-model, which shows good generalization with respect to different individuals, backgrounds, and attire.  ...  Kumar et al. (2010) use their OCMs to guide a probabilistic foreground segmentation scheme that shows acceptable performance in cutting out members of the encoded categories from cluttered images.  ... 
doi:10.1016/j.cogsys.2017.08.001 fatcat:yfzw4dcddje5tjb4k2f6rny7hq
« Previous Showing results 1 — 15 out of 469 results