Filters








502 Hits in 1.9 sec

Efficient Spatialtemporal Context Modeling for Action Recognition [article]

Congqi Cao, Yue Lu, Yifan Zhang, Dongmei Jiang, Yanning Zhang
2021 arXiv   pre-print
video for action recognition.  ...  However, directly modeling the contextual information between any two points brings huge cost in computation and memory, especially for action recognition, where there is an additional temporal dimension  ...  Conclusion In this paper, we have presented an efficient long-range context learning model for action recognition.  ... 
arXiv:2103.11190v2 fatcat:wrlnsswa7fcf5krgppgjukxhcm

Action Recognition using HighLevel Action Units

Paul T
2015 International Journal on Recent and Innovation Trends in Computing and Communication  
Testing phase uses the learned model for action prediction.Three components are used to classify activities such as New spatialtemporal descriptor, Statistics of the context-aware descriptors, Suppress  ...  In this project, a model is developed for human activity detection using high-level action units to represent human activity. Training phase learns the model for action units and action classifiers.  ...  This is more efficient in than BoVW models where hundreds of interest points are needed.  ... 
doi:10.17762/ijritcc2321-8169.150420 fatcat:rhgbdbolazacdgdnadui2wt6bq

EAN: Event Adaptive Network for Enhanced Action Recognition [article]

Yuan Tian, Yichao Yan, Xiongkuo Min, Guo Lu, Guangtao Zhai, Guodong Guo, Zhiyong Gao
2021 arXiv   pre-print
Efficiently modeling spatial-temporal information in videos is crucial for action recognition.  ...  On the other hand, the dense interaction modeling paradigm only achieves sub-optimal performance as action-irrelevant parts bring additional noises for the final prediction.  ...  In other words, the proposed LMC module is 1823× more efficient than optical flow modality, while achieving better performance for the action recognition task.  ... 
arXiv:2107.10771v1 fatcat:ywxhs7sbw5habbaokadhq6lpdy

Higher Order Recurrent Space-Time Transformer for Video Action Prediction [article]

Tsung-Ming Tai, Giuseppe Fiameni, Cheng-Kuang Lee, Oswald Lanz
2021 arXiv   pre-print
HORST achieves state of the art competitive performance on Something-Something early action recognition and EPIC-Kitchens action anticipation, showing evidence of predictive capability that we attribute  ...  The predominant modeling paradigm for this is sequence learning, mostly implemented through LSTMs.  ...  A.1 Early Action Recognition For early action recognition on SSv2 dataset, only the single action label for each sample is available.  ... 
arXiv:2104.08665v3 fatcat:skqeg4ryebg3zdert3txbw6yxy

Focusing and Diffusion: Bidirectional Attentive Graph Convolutional Networks for Skeleton-based Action Recognition [article]

Jialin Gao, Tong He, Xi Zhou, Shiming Ge
2019 arXiv   pre-print
recognition.  ...  A collection of approaches based on graph convolutional networks have proven success in skeleton-based action recognition by exploring neighborhood information and dense dependencies between intra-frame  ...  Introduction Action recognition is a challenging vision problem that has been studied for years.  ... 
arXiv:1912.11521v1 fatcat:g3budozjxjgbtl7eno6rpjjgni

Modeling Scene and Object Contexts for Human Action Retrieval With Few Examples

Yu-Gang Jiang, Zhenguo Li, Shih-Fu Chang
2011 IEEE transactions on circuits and systems for video technology (Print)  
For instance, driving car usually happens outdoors, and kissing involves two people moving toward each other. In this paper, we investigate the problem of context modeling for human action retrieval.  ...  One important contribution of this paper is to show that modeling the dependencies among actions, objects, and scenes can be efficiently achieved with very few examples.  ...  To the best of our knowledge, this is the first work on efficient modeling of action context using limited examples.  ... 
doi:10.1109/tcsvt.2011.2129870 fatcat:2ipnqennzneozoi2uhmd3lj3p4

Spatio-temporal Aware Non-negative Component Representation for Action Recognition [article]

Jianhong Wang, Tian Lan, Xu Zhang, Limin Luo
2016 arXiv   pre-print
The experimental results demonstrate the effectiveness of STANNCR for action recognition.  ...  This paper presents a novel mid-level representation for action recognition, named spatio-temporal aware non-negative component representation (STANNCR).  ...  result for action recognition.  ... 
arXiv:1608.07664v1 fatcat:nly55wzvj5c4lfmrjc2e7zdzh4

Disentangling and Unifying Graph Convolutions for Skeleton-Based Action Recognition [article]

Ziyu Liu, Hongwen Zhang, Zhenghao Chen, Zhiyong Wang, Wanli Ouyang
2020 arXiv   pre-print
Spatial-temporal graphs have been widely used by skeleton-based action recognition algorithms to model human action dynamics.  ...  To capture robust movement patterns from these graphs, long-range and multi-scale context aggregation and spatial-temporal dependency modeling are critical aspects of a powerful feature extractor.  ...  Skeleton-Based Action Recognition Earlier approaches [42, 6, 31, 36, 43, 48, 54] to skeletonbased action recognition focus on hand-crafting features and joint relationships for downstream classifiers  ... 
arXiv:2003.14111v2 fatcat:3tz5oc3jwbalto2rjxjpybiqpe

Adaptive Recursive Circle Framework for Fine-grained Action Recognition [article]

Hanxi Lin, Xinxiao Wu, Jiebo Luo
2021 arXiv   pre-print
How to model fine-grained spatial-temporal dynamics in videos has been a challenging problem for action recognition.  ...  We hope the ARC framework can facilitate fine-grained action recognition by introducing deeply refined features and multi-scale receptive fields at a low cost.  ...  In the context of fine-grained action recognition, rich and deep features are crucial for drawing conclusions on fine-grained details.  ... 
arXiv:2107.11813v1 fatcat:7drovvsuobfbdhgbtzv4hxo3fa

Human Activity Recognition based on Dynamic Spatio-Temporal Relations [article]

Zhenyu Liu, Yaqiang Yao, Yan Liu, Yuening Zhu, Zhenchao Tao, Lei Wang, Yuhong Feng
2020 arXiv   pre-print
In this paper, we develop a method for human activity recognition that tackles these two issues.  ...  Therefore, the description of a single human action and the modeling of the evolution of successive human actions are two major issues in human activity recognition.  ...  Conclusion In this paper, we developed an efficient method for human activity recognition from the perspective of human action modeling and its dynamical evolution.  ... 
arXiv:2006.16132v1 fatcat:snvksj7g5ra35d5bgj3vx2vm3q

Gate-Shift Networks for Video Action Recognition [article]

Swathikiran Sudhakaran, Sergio Escalera, Oswald Lanz
2020 arXiv   pre-print
Deep 3D CNNs for video action recognition are designed to learn powerful representations in the joint spatio-temporal feature space.  ...  We perform an extensive evaluation of the proposed module to study its effectiveness in video action recognition, achieving state-of-the-art results on Something Something-V1 and Diving48 datasets, and  ...  GST [27] makes use of the idea of grouped convolutions for developing an efficient architecture for action recognition.  ... 
arXiv:1912.00381v2 fatcat:5yujdzx7k5dcnliv3q5oxypzqe

Perceptron Synthesis Network: Rethinking the Action Scale Variances in Videos [article]

Yuan Tian, Guangtao Zhai, Zhiyong Gao
2022 arXiv   pre-print
Video action recognition has been partially addressed by the CNNs stacking of fixed-size 3D kernels.  ...  However, these methods may under-perform for only capturing rigid spatial-temporal patterns in single-scale spaces, while neglecting the scale variances across different action primitives.  ...  End-to-end learning for action recognition. Finally, we apply the proposed networks to the action recognition task.  ... 
arXiv:2007.11460v3 fatcat:q4ebmaqcxnetjnaqxp464vbd6i

SAFCAR: Structured Attention Fusion for Compositional Action Recognition [article]

Tae Soo Kim, Gregory D. Hager
2020 arXiv   pre-print
We present a general framework for compositional action recognition -- i.e. action recognition where the labels are composed out of simpler components such as subjects, atomic-actions and objects.  ...  The main challenge in compositional action recognition is that there is a combinatorially large set of possible actions that can be composed using basic components.  ...  Acknowledgement We thank Jin Bai for his consultations on object detectors and tracking . We thank Jonathan Jones for useful discussions.  ... 
arXiv:2012.02109v2 fatcat:wvqdumgwqbeivi7w3g24q7fbue

Action Keypoint Network for Efficient Video Recognition [article]

Xu Chen, Yahong Han, Xiaohan Wang, Yifan Sun, Yi Yang
2022 arXiv   pre-print
Reducing redundancy is crucial for improving the efficiency of video recognition models.  ...  Consequentially, AK-Net brings two-fold benefits for efficiency: The keypoint selection step collects informative content within arbitrary shapes and increases the efficiency for modeling spatial-temporal  ...  With these designs, AK-Net implements the video recognition tasks in a point cloud classification manner. AK-Net has two-fold advantages for efficient action recognition.  ... 
arXiv:2201.06304v1 fatcat:ckscf564zbckjdhff5u4inghr4

Dynamic Eye Movement Datasets and Learnt Saliency Models for Visual Action Recognition [chapter]

Stefan Mathe, Cristian Sminchisescu
2012 Lecture Notes in Computer Science  
Systems based on bag-of-words models operating on image features collected at maxima of sparse interest point operators have been extremely successful for both computer-based visual object and action recognition  ...  recognition task.  ...  vision algorithms in the context of action recognition have not been yet explored.  ... 
doi:10.1007/978-3-642-33709-3_60 fatcat:cznefptuejdgbgl7w2hawmq7qy
« Previous Showing results 1 — 15 out of 502 results