Filters








696 Hits in 4.5 sec

Fusing Image and Segmentation Cues for Skeleton Extraction in the Wild

Xiaolong Liu, Pengyuan Lyu, Xiang Bai, Ming-Ming Cheng
2017 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)  
Extracting skeletons from natural images is a challenging problem, due to complex backgrounds in the scene and various scales of objects.  ...  We find that the semantic segmentation probability map is complementary to the corresponding color image and can boost the performance of our baseline model which trained only on color images.  ...  To simplify the complex groundtruth fitting problem, we train a model which learns to combine image and segmentation cues for skeleton extraction.  ... 
doi:10.1109/iccvw.2017.205 dblp:conf/iccvw/LiuLBC17 fatcat:q7wzv6xwvvfwhpjlwk5bi7qghe

DeepFlux for Skeletons in the Wild [article]

Yukang Wang, Yongchao Xu, Stavros Tsogkas, Xiang Bai, Sven Dickinson, Kaleem Siddiqi
2018 arXiv   pre-print
Computing object skeletons in natural images is challenging, owing to large variations in object appearance and scale, and the complexity of handling background clutter.  ...  We evaluate the proposed method on three benchmark datasets for skeleton detection and two for symmetry detection, achieving consistently superior performance over state-of-the-art methods.  ...  [24] develop a two-stream network that combines image and segmentation cues to capture complementary information for skeleton localization.  ... 
arXiv:1811.12608v1 fatcat:pahjsjqtfbhzbmaawn4yd3ktxq

DeepSkeleton: Skeleton Map for 3D Human Pose Regression [article]

Qingfu Wan, Wei Zhang, Xiangyang Xue
2017 arXiv   pre-print
The effectiveness of our approach is validated on challenging in-the-wild dataset MPII and indoor dataset Human3.6M.  ...  For the first time, we show that training regression network from skeleton map alone is capable of meeting the performance of state-of-theart 3D human pose estimation works.  ...  The network structure shown in Figure 2 starts with a 224×224 image and extracts features along the downsampling process. Herein only res2c, res3d, res4f and res5c are sketched for brevity.  ... 
arXiv:1711.10796v1 fatcat:evdz5vymj5g4dpwfocdwhufm2i

Graph Neural Networks for Image Understanding Based on Multiple Cues: Group Emotion Recognition and Event Recognition as Use Cases [article]

Xin Guo, Luisa F. Polania, Bin Zhu, Charles Boncelet, Kenneth E. Barner
2020 arXiv   pre-print
A graph neural network (GNN) for image understanding based on multiple cues is proposed in this paper.  ...  Two image understanding tasks, namely group-level emotion recognition (GER) and event recognition, which are highly semantic and require the interaction of several deep models to synthesize multiple cues  ...  Acknowledgements The work is supported by the National Science Foundation under Grant No. 1319598.  ... 
arXiv:1909.12911v2 fatcat:qpaxwqgmwjhc3nqwah7bxtvd2y

Graph Neural Networks for Image Understanding Based on Multiple Cues: Group Emotion Recognition and Event Recognition as Use Cases

Xin Guo, Luisa F. Polania, Bin Zhu, Charles Boncelet, Kenneth E. Barner
2020 2020 IEEE Winter Conference on Applications of Computer Vision (WACV)  
A graph neural network (GNN) for image understanding based on multiple cues is proposed in this paper.  ...  Two image understanding tasks, namely group-level emotion recognition (GER) and event recognition, which are highly semantic and require the interaction of several deep models to synthesize multiple cues  ...  OpenPose [6, 42, 51] is used to extract skeleton images in the same way as described in [20, 22] . The SE-ResNet-50 is fine-tuned on skeleton images in the same way as described in [22] .  ... 
doi:10.1109/wacv45572.2020.9093547 dblp:conf/wacv/GuoPZBB20 fatcat:lzz7pps6fvhgjatqe4d76kjcae

SRN: Side-Output Residual Network for Object Symmetry Detection in the Wild

Wei Ke, Jie Chen, Jianbin Jiao, Guoying Zhao, Qixiang Ye
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
symmetry detection in the wild.  ...  SK506 Figure 1: We propose a new benchmark, named Sym-PASCAL, for object symmetry detection in the wild.  ...  Acknowledgement This work is supported in partial by the NSFC under  ... 
doi:10.1109/cvpr.2017.40 dblp:conf/cvpr/KeCJZY17 fatcat:e7upwfjwmbe73cqlvsthw26jhe

SRN: Side-output Residual Network for Object Symmetry Detection in the Wild [article]

Wei Ke, Jie Chen, Jianbin Jiao, Guoying Zhao, Qixiang Ye
2017 arXiv   pre-print
symmetry detection in the wild.  ...  In this paper, we establish a baseline for object symmetry detection in complex backgrounds by presenting a new benchmark and an end-to-end deep learning approach, opening up a promising direction for  ...  Acknowledgement This work is partially supported by NSFC under Grant 61671427, Beijing Municipal Science and Technology Commission under Grant Z161100001616005, and Science and Technology Innovation Foundation  ... 
arXiv:1703.02243v2 fatcat:qgrtemx3a5hdrbyr6e7v64hqhq

RSRN: Rich Side-Output Residual Network for Medial Axis Detection

Chang Liu, Wei Ke, Jianbin Jiao, Qixiang Ye
2017 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)  
In this paper, we propose a Rich Side-output Residual Network (RSRN) for medial axis detection for the ICCV 2017 workshop challenge on detecting symmetry in the wild.  ...  RSRN uses the rich features of fully convolutional network by hierarchically fusing side-outputs in a deep-toshallow manner to decrease the residual between the detection result and the ground-truth, which  ...  Acknowledgement This work is partially supported by NSFC under Grant 61671427 and Grant 61771447.  ... 
doi:10.1109/iccvw.2017.204 dblp:conf/iccvw/LiuKJY17 fatcat:jbi4d2wz7fawzpzhhnsw3iv4xa

Beyond Planar Symmetry: Modeling human perception of reflection and rotation symmetries in the wild [article]

Christopher Funk, Yanxi Liu
2017 arXiv   pre-print
Motivated by a new study demonstrating the extremely high inter-person accuracy of human perceived symmetries in the wild, we have constructed the first deep-learning neural network for reflection and  ...  We employ novel methods to convert discrete human labels into symmetry heatmaps, capture symmetry densely in an image and quantitatively evaluate Sym-NET against multiple existing computer vision algorithms  ...  [59] use a deep CNN to learn symmetry at multiple scales and fuse the final output together. The network needs object skeleton ground-truth for the particular scale of the objects.  ... 
arXiv:1704.03568v2 fatcat:hzg7p2d6sjfsdb5nutrdf4rfmy

SRN: Side-output Residual Network for Object Reflection Symmetry Detection and Beyond [article]

Wei Ke and Jie Chen and Jianbin Jiao and Guoying Zhao and Qixiang Ye
2019 arXiv   pre-print
for symmetry detection in the wild.  ...  In this paper, we establish a baseline for object reflection symmetry detection in complex backgrounds by presenting a new benchmark and an end-to-end deep learning approach, opening up a promising direction  ...  ACKNOWLEDGMENTS The authors would like to express their sincere appreciation to Prof. David Doermann and Dr. Pengxu Wei for their comments and suggestions.  ... 
arXiv:1807.06621v2 fatcat:qhrzqziwyrhg3p64wifyhbtt6u

RGB-D-based Human Motion Recognition with Deep Learning: A Survey [article]

Pichao Wang and Wanqing Li and Philip Ogunbona and Jun Wan and Sergio Escalera
2018 arXiv   pre-print
In particular, convolutional neural networks (CNN) have achieved great success for image-based tasks, and recurrent neural networks (RNN) are renowned for sequence-based problems.  ...  The reviewed methods are broadly categorized into four groups, depending on the modality adopted for recognition: RGB-based, depth-based, skeleton-based and RGB+D-based.  ...  [151] took a different approach and adopted scene flow to extract features that fused the RGB and depth from the onset.  ... 
arXiv:1711.08362v2 fatcat:cugugpqeffcshnwwto4z2aw4ti

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
Motivated by this insight, we propose DIGIT, a novel method for estimating the 3D poses of two interacting hands from a single monocular image.  ...  In contrast to prior work, we do not decouple the segmentation from the pose estimation stage, but rather leverage the per-pixel probabilities directly in the downstream pose estimation task.  ...  We thank Korrawe Karunratanakul, Emre Aksan, and Dimitrios Tzionas for their feedback. Disclosure. MJB has received research gift funds from Adobe, Intel, Nvidia, Facebook, and Amazon.  ... 
arXiv:2107.00434v2 fatcat:frrzchccbja4pclwkh4ebhposi

Characterness: An Indicator of Text in the Wild

Yao Li, Wenjing Jia, Chunhua Shen, Anton van den Hengel
2014 IEEE Transactions on Image Processing  
In order to measure the characterness we develop three novel cues that are tailored for character detection, and a Bayesian method for their integration.  ...  Text in an image provides vital information for interpreting its contents, and text in a scene can aide with a variety of tasks from navigation, to obstacle avoidance, and odometry.  ...  ACKNOWLEDGMENT This work is in part supported by ARC Grants FT120100969 and LP130100156; the UTS FEIT Industry and Innovation Project Scheme.  ... 
doi:10.1109/tip.2014.2302896 pmid:24808338 fatcat:lonk5aciqzf3tjb72fjb6p2c6u

Characterness: An Indicator of Text in the Wild [article]

Yao Li, Wenjing Jia, Chunhua Shen, Anton van den Hengel
2013 arXiv   pre-print
In order to measure the characterness we develop three novel cues that are tailored for character detection, and a Bayesian method for their integration.  ...  Text in an image provides vital information for interpreting its contents, and text in a scene can aide with a variety of tasks from navigation, to obstacle avoidance, and odometry.  ...  ACKNOWLEDGMENT This work is in part supported by ARC Grants FT120100969 and LP130100156; the UTS FEIT Industry and Innovation Project Scheme.  ... 
arXiv:1309.6691v1 fatcat:dktsea3psvhvxayjnh247w74ku

Unsupervised 3D Pose Estimation with Geometric Self-Supervision [article]

Ching-Hang Chen, Ambrish Tyagi, Amit Agrawal, Dylan Drover, Rohith MV, Stefan Stojanov, James M. Rehg
2019 arXiv   pre-print
Additionally, to learn from 2D poses "in the wild", we train an unsupervised 2D domain adapter network to allow for an expansion of 2D data.  ...  By lifting the synthetic 2D poses back to 3D and re-projecting them in the original camera view, we can define self-consistency loss both in 3D and in 2D.  ...  Exemplar based methods [5, 19, 51] use a database/dictionary of 3D skeletons for nearest-neighbor look-up. Tekin et al. [42] fused 2D and 3D image cues re-lying on 2D-3D correspondences.  ... 
arXiv:1904.04812v1 fatcat:4scdgh64ffavzcevewwmvo3he4
« Previous Showing results 1 — 15 out of 696 results