138,110 Hits in 5.4 sec

Learning Class Regularized Features for Action Recognition [article]

Alexandros Stergiou, Ronald Poppe, Remco C. Veltkamp
2020 arXiv   pre-print
We show that using Class Regularization blocks in state-of-the-art CNN architectures for action recognition leads to systematic improvement gains of 1.8%, 1.2% and 1.4% on the Kinetics, UCF-101 and HMDB  ...  To overcome the class-agnostic use of kernels in CNNs, we introduce a novel method named Class Regularization that performs class-based regularization of layer activations.  ...  features. • We report performance gains for benchmark action recognition datasets Kinetics, UCF-101 and HMDB-51 by including Class Regularization blocks.  ... 
arXiv:2002.02651v1 fatcat:amxchvqt2fgrxg2sitdkszxee4

Adversarial Self-Supervised Learning for Semi-Supervised 3D Action Recognition [article]

Chenyang Si, Xuecheng Nie, Wei Wang, Liang Wang, Tieniu Tan, Jiashi Feng
2020 arXiv   pre-print
However, few effective self-supervised approaches exist for 3D action recognition, and directly applying SSL for semi-supervised learning suffers from misalignment of representations learned from SSL and  ...  We further propose an adversarial regularization to align the feature distributions of labeled and unlabeled samples.  ...  The ASSL leverages SSL and adversarial regularization for semi-supervised 3D action recognition.  ... 
arXiv:2007.05934v1 fatcat:5wfusiu7zzcsxdclrbonzkpqxm

Multimodal Multipart Learning for Action Recognition in Depth Videos [article]

Amir Shahroudy, Gang Wang, Tian-Tsong Ng, Qingxiong Yang
2015 arXiv   pre-print
The articulated and complex nature of human actions makes the task of action recognition difficult.  ...  The proper structure of multimodal multipart features are formulated into the learning framework via the proposed hierarchical mixed norm, to regularize the structured features of each part and to apply  ...  for action recognition.  ... 
arXiv:1507.08761v1 fatcat:3c2ifw37c5h2jaanwx4lix3ajy

Multiview Hessian regularized logistic regression for action recognition [article]

W. Liu, H. Liu, D. Tao, Y. Wang, Ke Lu
2014 arXiv   pre-print
logistic regression for human action recognition.  ...  To tackle the above problems, in this paper we propose multiview Hessian regularized logistic regression (mHLR) for human action recognition.  ...  Therefore, in this paper we propose multiview Hessian regularized logistic regression (mHLR) for human action recognition.  ... 
arXiv:1403.0829v1 fatcat:dhnbx4goxjdw7jqjqbj7gpfoxa

Graph regularized implicit pose for 3D human action recognition

Tommi Kerola, Nakamasa Inoue, Koichi Shinoda
2016 2016 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA)  
We present a novel feature descriptor for 3D human action recognition using graph signal processing techniques.  ...  The graph total variation regularizer learns an action representation that encourages piece-wise constantness, which helps discriminating between different action classes.  ...  In this paper, we illustrate the advantage of graph regularizers for learning a linear subspace embedding suitable for KNN-based action recognition.  ... 
doi:10.1109/apsipa.2016.7820717 dblp:conf/apsipa/KerolaIS16 fatcat:66he7tvgtrbz5dfzz4zbdxtvyq

Regularizing Long Short Term Memory with 3D Human-Skeleton Sequences for Action Recognition

Behrooz Mahasseni, Sinisa Todorovic
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
poorly represented or missing features of human actions in the training videos.  ...  For recognition, we use Long Short Term Memory (LSTM) grounded via a deep Convolutional Neural Network (CNN) onto the video.  ...  Regularizing LSTM for Action Recognition As mentioned in Sec. 1, our novel architecture consists of eLSTM for learning a feature representation of 3D human-skeleton sequences, and a stacked DCNN+LSTM for  ... 
doi:10.1109/cvpr.2016.333 dblp:conf/cvpr/MahasseniT16 fatcat:rf3w6jyywfaz7ibf6uqjhso4hu

Supervised Local Descriptor Learning for Human Action Recognition

Xiantong Zhen, Feng Zheng, Ling Shao, Xianbin Cao, Dan Xu
2017 IEEE transactions on multimedia  
local feature descriptors for more accurate and efficient action recognition.  ...  The proposed I2CDDE for the first time achieves fully supervised learning of local feature descriptors.  ...  local feature descriptors in the video domain for action recognition.  ... 
doi:10.1109/tmm.2017.2700204 fatcat:eztuscrzencolowefxomah4s7u

Improved Collaborative Representation Classifier Based on l2-Regularized for Human Action Recognition

Shirui Huo, Tianrui Hu, Ce Li
2017 Journal of Electrical and Computer Engineering  
The improved collaborative representation classifier (ICRC) based on l2-regularized for human action recognition is presented to maximize the likelihood that a test sample belongs to each class, then theoretical  ...  Human action recognition is an important recent challenging task.  ...  ICRC for action recognition.  ... 
doi:10.1155/2017/8191537 fatcat:z7tg3adle5dbhgz2za3aaeh6ra

Class-Incremental Learning for Action Recognition in Videos [article]

Jaeyoo Park, Minsoo Kang, Bohyung Han
2022 arXiv   pre-print
We tackle catastrophic forgetting problem in the context of class-incremental learning for video recognition, which has not been explored actively despite the popularity of continual learning.  ...  We evaluate the proposed approach on brand-new splits of class-incremental action recognition benchmarks constructed upon the UCF101, HMDB51, and Something-Something V2 datasets, and demonstrate the effectiveness  ...  learning for action recognition based on temporally attentive knowledge distillation.  ... 
arXiv:2203.13611v1 fatcat:f436lu5mhvhirgze5h3ncp5gpi

Learning Representations from Spatio-Temporal Distance Maps for 3D Action Recognition with Convolutional Neural Networks

2020 Advances in Distributed Computing and Artificial Intelligence Journal  
., 2017) and two stream RNN (Wang and Wang, 2017) have been proposed to learn discriminative features from skeleton data for action recognition.  ...  for action recognition.  ...  discriminative features from skeleton data for action recognition.  ... 
doi:10.14201/adcai201982518 fatcat:v37tvpica5b63kym2ozobq436m

Deep Spatio-temporal Manifold Network for Action Recognition [article]

Ce Li, Chen Chen, Baochang Zhang, Qixiang Ye, Jungong Han, Rongrong Ji
2017 arXiv   pre-print
We propose leveraging the manifold structure to constrain the deep action feature learning, thereby minimizing the intra-class variations in the feature space and alleviating the over-fitting problem.  ...  Considering that manifold can be transferred, layer by layer, from the data domain to the deep features, the manifold priori is posed from the top layer into the back propagation learning procedure of  ...  Experimental results on two benchmark datasets demonstrated that our STMN method achieves competitive results for human action results.  ... 
arXiv:1705.03148v1 fatcat:fycshjvvcbhlhhz5scvyrkpx3a

A Generative Approach to Zero-Shot and Few-Shot Action Recognition [article]

Ashish Mishra, Vinay Kumar Verma, M Shiva Krishna Reddy, Arulkumar S, Piyush Rai, Anurag Mittal
2018 arXiv   pre-print
We present a generative framework for zero-shot action recognition where some of the possible action classes do not occur in the training data.  ...  Our framework also naturally extends to few-shot action recognition where a few labeled examples from unseen classes are available.  ...  Hema A Murthy for her valuable feedback. Vinay Verma acknowledges support from Visvesvaraya Ph.D. fellowship.  ... 
arXiv:1801.09086v1 fatcat:m5pdvxu57fd47or3lraqdlnsze

Learning Cross-modal Contrastive Features for Video Domain Adaptation [article]

Donghyun Kim, Yi-Hsuan Tsai, Bingbing Zhuang, Xiang Yu, Stan Sclaroff, Kate Saenko, Manmohan Chandraker
2021 arXiv   pre-print
Learning transferable and domain adaptive feature representations from videos is important for video-relevant tasks such as action recognition.  ...  To this end, we propose a unified framework for video domain adaptation, which simultaneously regularizes cross-modal and cross-domain feature representations.  ...  Domain Adaptation for Action Recognition.  ... 
arXiv:2108.11974v1 fatcat:o6el6nmcpfeclninzwdobezw4y

Handcrafted vs. learned representations for human action recognition

Xiantong Zhen, Ling Shao, Stephen J. Maybank, Rama Chellappa
2016 Image and Vision Computing  
Acknowledgement We would like to thank all the authors for their contributions to this special issue, and reviewers for their timely and insightful reviews. We thank Professors J.-M. Frahm and M.  ...  Pantic, Editor-in-Chief of the Image and Vision Computing Journal for giving us the opportunity to guest edit this special issue, and the Elsevier staff, Yanhong Zhai for her great support to this special  ...  of feature learning/engineering for human action recognition.  ... 
doi:10.1016/j.imavis.2016.10.002 fatcat:j4c2txj3g5glra67qvzab5mmke

Transductive Zero-Shot Action Recognition by Word-Vector Embedding [article]

Xun Xu, Timothy Hospedales, Shaogang Gong
2016 arXiv   pre-print
The number of categories for action recognition is growing rapidly and it has become increasingly hard to label sufficient training data for learning conventional models for all categories.  ...  In this work, we explore word-vectors as the shared semantic space to embed videos and category labels for ZSL action recognition.  ...  ZSL for Action Recognition Despite clear appeal from ZSL, few studies have considered it for action recognition.  ... 
arXiv:1511.04458v2 fatcat:yxfn52pdhjfatedmixz4evtiay
« Previous Showing results 1 — 15 out of 138,110 results