A Unified Multimodal De- and Re-Coupling Framework for RGB-D Motion Recognition
Open accessThe DOI matches its Crossref record; 86 of the 86 references it cites were found in a registry or an index.
Abstract
Motion recognition is a promising direction in computer vision, but the training of video classification models is much harder than images due to insufficient data and considerable parameters. To get around this, some works strive to explore multimodal cues from RGB-D data. Although improving motion recognition to some extent, these methods still face sub-optimal situations in the following aspects: (i) Data augmentation, i.e., the scale of the RGB-D datasets is still limited, and few efforts have been made to explore novel data augmentation strategies for videos; (ii) Optimization mechanism, i.e., the tightly space-time-entangled network structure brings more challenges to spatiotemporal information modeling; And (iii) cross-modal knowledge fusion, i.e., the high similarity between multimodal representations leads to insufficient late fusion. To alleviate these drawbacks, we propose to improve RGB-D-based motion recognition both from data and algorithm perspectives in this article. In more detail, firstly, we introduce a novel video data augmentation method dubbed ShuffleMix, which acts as a supplement to MixUp, to provide additional temporal regularization for motion recognition. Secondly, a Unified Multimodal De-coupling and multi-stage Re-coupling framework, termed UMDR, is proposed for video representation learning. Finally, a novel cross-modal Complement Feature Catcher (CFCer) is explored to mine potential commonalities features in multimodal information as the auxiliary fusion stream, to improve the late fusion results. The seamless combination of these novel designs forms a robust spatiotemporal representation and achieves better performance than state-of-the-art methods on four public motion datasets. Specifically, UMDR achieves unprecedented improvements of ↑ 4.5% on the Chalearn IsoGD dataset.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“However, existing works on missing modality robustness vary widely in terms of model architectures [32, 4], fusion methods [35, 63], training procedures [20, 32], and missing feature generation methods [23]. Due to this heterogeneity, a fair comparison based solely on model size/number of parameters is infeasible.”
From Robust Multimodal Learning with Missing Modalities via Parameter-Efficient Adaptation (2025) · cites this paper as [63]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“As shown in Table 5, our adaptation performs better compared to recent modality masking and generation based approach ActionMAE [23] and modality de- and re-coupling based approaches Motion-RGBD [91] and UMDR [63].”
From Robust Multimodal Learning with Missing Modalities via Parameter-Efficient Adaptation (2025) · cites this paper as [63]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 47,472 · cited 1×
“To achieve the goal in the former aspect, inspired by self-attention [46], we calculate dynamic attention maps based on the input to re-weight spatial features of each frame to highlight sensitive filters in it.”
MethodDistilling the Knowledge in a Neural Network
Hinton, G., Vinyals, O., & Dean, J. (2015) · arXiv · doi:10.48550/arxiv.1503.02531 · Cited by 14,204
A survey on Image Data Augmentation for Deep Learning
Shorten, C., & Khoshgoftaar, T. M. (2019) · Journal of Big Data · doi:10.1186/s40537-019-0197-0 · Cited by 13,203 · cited 1×
“Data augmentation, a data-driven and informed regularization strategy that artificially increases the number of training samples to improve the generalization ability of models, has been widely explored for better training of deep neural networks [20], [21], [22], [23].”
BackgroundQuo Vadis, Action Recognition? A New Model and the Kinetics Dataset
Carreira, J., & Zisserman, A. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.502 · Cited by 9,570
mixup: Beyond Empirical Risk Minimization
Zhang, H., Cisse, M., Dauphin, Y. N., & Lopez-Paz, D. (2017) · arXiv · doi:10.48550/arxiv.1710.09412 · Cited by 5,831 · cited 3×
“The latter refers to utilizing multiple samples to generate a new instance, including video MixUp [14] and CutMix [28].”
BackgroundMethodCutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features
Yun, S., Han, D., Chun, S., Oh, S. J., Yoo, Y., & Choe, J. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00612 · Cited by 5,400 · cited 1×
“The latter refers to utilizing multiple samples to generate a new instance, including video MixUp [14] and CutMix [28].”
BackgroundSGDR: Stochastic Gradient Descent with Warm Restarts
Loshchilov, I., & Hutter, F. (2016) · arXiv · doi:10.48550/arxiv.1608.03983 · Cited by 5,309 · cited 1×
“The learning rate is linearly ramped up to 0.01 during the first 5 epochs and then decayed with a cosine schedule [63].”
BackgroundInfluentialInception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
Szegedy, C., Ioffe, S., Vanhoucke, V., & Alemi, A. (2017) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v31i1.11231 · Cited by 4,636 · cited 1×
“Similar to [45], the Steam module is the initial set of operations used to map low-level visual embedding before executing to the SMS module, written as: where Θ represents the learnable parameter matrices.”
MethodSlowFast Networks for Video Recognition
Feichtenhofer, C., Fan, H., Malik, J., & He, K. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00630 · Cited by 4,007 · cited 1×
“For each block, inspired by [11], [47], we construct multi-branch structures to tackle inputs with different temporal resolutions and feature dimensions accordingly: fast ( R ), moderate ( R T × C ), and slow ( R ), to perceive different motion variations, i.e., the high temporal resolution…”
MethodTemporal Segment Networks: Towards Good Practices for Deep Action Recognition
Wang, L., Xiong, Y., Wang, Z., Qiao, Y., Lin, D., Tang, X., & Van Gool, L. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46484-8_2 · Cited by 3,958
A Closer Look at Spatiotemporal Convolutions for Action Recognition
Tran, D., Wang, H., Torresani, L., Ray, J., LeCun, Y., & Paluri, M. (2017) · arXiv · doi:10.48550/arxiv.1711.11248 · Cited by 3,664 · cited 1×
“…information is hard to filter in the entangled space-time space, leading to decreased efficiency and robustness of the model; Although some works [15], [16], [17] attempt to decompose the 3D modeling procedure to decouple spatial and temporal learning, they are still problematic in terms of…”
BackgroundA Closer Look at Spatiotemporal Convolutions for Action Recognition
Tran, D., Wang, H., Torresani, L., Ray, J., LeCun, Y., & Paluri, M. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00675 · Cited by 3,516
NTU RGB+D: A Large Scale Dataset for 3D Human Activity Analysis
Shahroudy, A., Liu, J., Ng, T.-T., & Wang, G. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.115 · Cited by 3,148 · cited 3×
“We evaluate our method on four public RGB-D motion datasets: NTU RGB-D [56], NvGesture [57], Chalearn IsoGD [58], [59] and THU-READ [60] datasets.”
MethodA Tutorial on Principal Component Analysis
Shlens, J. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1404.1100 · Cited by 2,229 · cited 1×
“(a) Visualization of feature distribution dimensional descended by PCA [51].”
BackgroundBeyond short snippets: Deep networks for video classification
Ng, J. Y.-H., Hausknecht, M., Vijayanarasimhan, S., Vinyals, O., Monga, R., & Toderici, G. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7299101 · Cited by 2,062 · cited 1×
“[24] leverage an upsampling strategy to significantly improve the performance on human action recognition tasks.”
BackgroundMISA: Modality-Invariant and -Specific Representations for Multimodal Sentiment Analysis
Hazarika, D., Zimmermann, R., & Poria, S. (2020) · Proceedings of the 28th ACM International Conference on Multimedia · doi:10.1145/3394171.3413678 · Cited by 1,073 · cited 3×
“The NTU RGB-D is a large-scale human action dataset containing more than 56,000 multi-view RGB-D videos with 4 million frames that include 60 actions performed by 40 subjects.”
BackgroundMethodChannel-wise Topology Refinement Graph Convolution for Skeleton-Based Action Recognition
Chen, Y., Zhang, Z., Yuan, C., Li, B., Deng, Y., & Hu, W. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.01311 · Cited by 1,039
Skeleton-Based Action Recognition With Shift Graph Convolutional Network
Cheng, K., Zhang, Y., He, X., Chen, W., Cheng, J., & Lu, H. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00026 · Cited by 981 · cited 1×
“And the performance of the single modality even surpasses most skeleton-based approaches [67], [68], [69], which further demonstrates the robustness of our method to noisy backgrounds and its strong motion perception abilities.”
BackgroundSkeleton-Based Action Recognition With Directed Graph Neural Networks
Shi, L., Zhang, Y., Cheng, J., & Lu, H. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00810 · Cited by 930
Revisiting Skeleton-based Action Recognition
Duan, H., Zhao, Y., Chen, K., Lin, D., & Dai, B. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.00298 · Cited by 834 · cited 1×
“Furthermore, we also compare with other multimodal-based approaches such as Deep Bilinear [72], MMTM [10] and PoseC3D [74].”
ResultMolchanov, P., Yang, X., Gupta, S., Kim, K., Tyree, S., & Kautz, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.456 · Cited by 695 · cited 11×
“The NvGesture dataset focuses on touchless driver controlling.”
MethodInfluentialA Comprehensive Survey of Image Augmentation Techniques for Deep Learning
Xu, M., Yoon, S., Fuentes, A., & Park, D. S. (2023) · Pattern Recognition · doi:10.1016/j.patcog.2023.109347 · Cited by 634 · cited 1×
“Data augmentation, a data-driven and informed regularization strategy that artificially increases the number of training samples to improve the generalization ability of models, has been widely explored for better training of deep neural networks [20], [21], [22], [23].”
BackgroundModeling video evolution for action recognition
Fernando, B., Gavves, E., Jose Oramas, M., Ghodrati, A., & Tuytelaars, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7299176 · Cited by 569 · cited 1×
“Basura et al. [26] attempt to learn video-wide representation by arranging frames in chronological order.”
BackgroundChen, X., Lin, K.-Y., Wang, J., Wu, W., Qian, C., Li, H., & Zeng, G. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-58621-8_33 · Cited by 488
Decoupling GCN with DropGraph Module for Skeleton-Based Action Recognition
Cheng, K., Zhang, Y., Cao, C., Shi, L., Cheng, J., & Lu, H. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-58586-0_32 · Cited by 384 · cited 1×
“And the performance of the single modality even surpasses most skeleton-based approaches [67], [68], [69], which further demonstrates the robustness of our method to noisy backgrounds and its strong motion perception abilities.”
BackgroundEarly Convolutions Help Transformers See Better
Xiao, T., Singh, M., Mintun, E., Darrell, T. J., Dollár, P., & Girshick, R. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.14881 · Cited by 339
Real-time Hand Gesture Detection and Classification Using Convolutional Neural Networks
Kopuklu, O., Gunduz, A., Kose, N., & Rigoll, G. (2019) · 2019 14th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2019) · doi:10.1109/fg.2019.8756576 · Cited by 319
ChaLearn Looking at People RGB-D Isolated and Continuous Datasets for Gesture Recognition
Wan, J., Li, S. Z., Zhao, Y., Zhou, S., Guyon, I., & Escalera, S. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2016.100 · Cited by 276 · cited 1×
“We evaluate our method on four public RGB-D motion datasets: NTU RGB-D [56], NvGesture [57], Chalearn IsoGD [58], [59] and THU-READ [60] datasets.”
MethodSelf-citationMultimodal Gesture Recognition Using 3-D Convolution and Convolutional LSTM
Zhu, G., Zhang, L., Shen, P., & Song, J. (2017) · IEEE Access · doi:10.1109/access.2017.2684186 · Cited by 270 · cited 1×
“RGB-D-based motion recognition [8], [13], [19], [32], [33], [34], [35], [36], [37], [38], thus, has attracted extensive attention thanks to the availability of additional depth cues.”
BackgroundDeep Multimodal Feature Analysis for Action Recognition in RGB+D Videos
Shahroudy, A., Ng, T.-T., Gong, Y., & Wang, G. (2018) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2691321 · Cited by 264
The Jester Dataset: A Large-Scale Video Dataset of Human Gestures
Materzynska, J., Berger, G., Bax, I., & Memisevic, R. (2019) · 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW) · doi:10.1109/iccvw.2019.00349 · Cited by 260 · cited 1×
“Similar to [11], [13], [19], all of our experiments except NTU-RGBD are pre-trained on 20BN Jester V1 dataset [64].”
ResultMMTM: Multimodal Transfer Module for CNN Fusion
Vaezi Joze, H. R., Shaban, A., Iuzzolino, M. L., & Koishida, K. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.01330 · Cited by 259
Learning Spatiotemporal Features Using 3DCNN and Convolutional LSTM for Gesture Recognition
Zhang, L., Zhu, G., Shen, P., Song, J., Shah, S. A., & Bennamoun, M. (2017) · 2017 IEEE International Conference on Computer Vision Workshops (ICCVW) · doi:10.1109/iccvw.2017.369 · Cited by 241
How Does Mixup Help With Robustness and Generalization?Preprint
Zhang, L., Deng, Z., Kawaguchi, K., Ghorbani, A., & Zou, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.04819 · Cited by 240
Zolfaghari, M., Oliveira, G. L., Sedaghat, N., & Brox, T. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.316 · Cited by 235
Point 4D Transformer Networks for Spatio-Temporal Modeling in Point Cloud Videos
Fan, H., Yang, Y., & Kankanhalli, M. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.01398 · Cited by 210 · cited 1×
“7 P4Transformer [73] Point clouds - 90.”
KVT: k-NN Attention for Boosting Vision Transformers
Wang, P., Wang, X., Wang, F., Lin, M., Chang, S., Li, H., & Jin, R. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-20053-3_17 · Cited by 199
Self-citationImproving the Performance of Unimodal Dynamic Hand-Gesture Recognition With Multimodal Training
Abavisani, M., Joze, H. R. V., & Patel, V. M. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00126 · Cited by 190 · cited 2×
“H UMAN motion recognition [1], [2], [3] is an attractive yet challenging task.”
BackgroundDepth Pooling Based Large-Scale 3-D Action Recognition With Convolutional Neural Networks
Wang, P., Li, W., Gao, Z., Tang, C., & Ogunbona, P. O. (2018) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2018.2818329 · Cited by 182 · cited 1×
“Recently, RGB-D-based motion recognition [4], [5], [6] has attracted much attention in computer vision due to the easy availability of depth data and its extensive application prospects such as video surveillance and human-object interfaces.”
BackgroundSelf-citationTraining data-efficient image transformers & distillation through attention
Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., & Jégou, H. (2020) · arXiv · doi:10.48550/arxiv.2012.12877 · Cited by 182 · cited 1×
“Data augmentation, a data-driven and informed regularization strategy that artificially increases the number of training samples to improve the generalization ability of models, has been widely explored for better training of deep neural networks [20], [21], [22], [23].”
BackgroundMultimodal Gesture Recognition Based on the ResC3D Network
Miao, Q., Li, Y., Ouyang, W., Ma, Z., Xu, X., Shi, W., & Cao, X. (2017) · 2017 IEEE International Conference on Computer Vision Workshops (ICCVW) · doi:10.1109/iccvw.2017.360 · Cited by 168
Late Temporal Modeling in 3D CNN Architectures with BERT for Action Recognition
Kalfaoglu, M. E., Kalkan, S., & Alatan, A. A. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-68238-5_48 · Cited by 161
Gesture Recognition: Focus on the Hands
Narayana, P., Beveridge, J. R., & Draper, B. A. (2018) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2018.00549 · Cited by 156
Searching Multi-Rate and Multi-Modal Temporal Enhanced Networks for Gesture Recognition
Yu, Z., Zhou, B., Wan, J., Wang, P., Chen, H., Liu, X., Li, S. Z., & Zhao, G. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3087348 · Cited by 156 · cited 7×
“Furthermore, late fusion is still popular in RGB-D-based motion recognition [2], [11], [13], [18] due to its flexibility and effectiveness.”
BackgroundMethodResultInfluentialSelf-citationWang, P., Li, W., Gao, Z., Zhang, Y., Tang, C., & Ogunbona, P. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.52 · Cited by 151 · cited 2×
“In addition, scene flow is adopted in [34] for compact RGB-D representation learning.”
BackgroundMethodSelf-citationElboushaki, A., Hannane, R., Afdel, K., & Koutti, L. (2019) · Expert Systems with Applications · doi:10.1016/j.eswa.2019.112829 · Cited by 151
StNet: Local and Global Spatial-Temporal Modeling for Action Recognition
He, D., Zhou, Z., Gan, C., Li, F., Liu, X., Li, Y., Wang, L., & Wen, S. (2019) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v33i01.33018401 · Cited by 143
The Effects of Regularization and Data Augmentation Are Class Dependent
Balestriero, R., Bottou, L., & LeCun, Y. (2022) · neural information processing systems · doi:10.52202/068431-2745 · Cited by 131
Wang, J., Gao, Y., Li, K., Lin, Y., Ma, A. J., Cheng, H., Peng, P., Huang, F., Ji, R., & Sun, X. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.01163 · Cited by 128 · cited 1×
“[30] selects a static frame and adds it to every other frame to construct a distracting video, thereby alleviating the problem of implicit biases that commonly exists in video datasets.”
BackgroundTrear: Transformer-Based RGB-D Egocentric Action Recognition
Li, X., Hou, Y., Wang, P., Gao, Z., Xu, M., & Li, W. (2021) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2020.3048883 · Cited by 125 · cited 2×
“At present, although various architectures [10], [11], [12], [13] have been continuously suggested to improve the multimodal motion recognition (as shown in Fig.”
BackgroundSelf-citationDeep Bilinear Learning for RGB-D Action Recognition
Hu, J., Zheng, W., Pan, J., Lai, J., & Zhang, J. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01234-2_21 · Cited by 110 · cited 1×
“Furthermore, we also compare with other multimodal-based approaches such as Deep Bilinear [72], MMTM [10] and PoseC3D [74].”
ResultCooperative Training of Deep Aggregation Networks for RGB-D Action Recognition
Wang, P., Li, W., Wan, J., Ogunbona, P., & Liu, X. (2018) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v32i1.12228 · Cited by 107
Self-citationVideoMix: Rethinking Data Augmentation for Video ClassificationPreprint
Yun, S., Oh, S. J., Heo, B., Han, D., & Kim, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.03457 · Cited by 105 · cited 1×
“Finally, we improve the multimodal late fusion by exploring the modality-specific complementary features (Section III-C).”
MethodExplore Efficient Local Features from RGB-D Data for One-Shot Learning Gesture Recognition
Wan, J., Guo, G., & Li, S. Z. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2513479 · Cited by 103 · cited 1×
“RGB-D-based motion recognition refers to jointly exploring the color and depth cues in multimodal data by constructing a multi-stream processing mechanism, thereby forming an informative joint representation for motion description [7], [8], [9].”
BackgroundSelf-citationOn Mixup RegularizationPreprint
Carratino, L., Cissé, M., Jenatton, R., & Vert, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.06049 · Cited by 101 · cited 2×
“Taking the cross-entropy loss l ( · , · ) as an example, the loss function based on MixUp can be described as (refer to [43] for more): where , which are given by: where Furthermore, δ i and (cid:4) i represent random spatial perturbations of input and output, respectively.”
MethodRedundancy and Attention in Convolutional LSTM for Gesture Recognition
Zhu, G., Zhang, L., Yang, L., Mei, L., Shah, S. A. A., Bennamoun, M., & Shen, P. (2019) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2019.2919764 · Cited by 97 · cited 1×
“Furthermore, late fusion is still popular in RGB-D-based motion recognition [2], [11], [13], [18] due to its flexibility and effectiveness.”
BackgroundMulti-Modal Face Anti-Spoofing Based on Central Difference Networks
Yu, Z., Qin, Y., Li, X., Wang, Z., Zhao, C., Lei, Z., & Zhao, G. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw50498.2020.00333 · Cited by 96 · cited 1×
“Similar to most of the recent works [13], [19], [62], [63], we also evaluate the experimental results on the validation set for this dataset.”
ResultImproving Human Action Recognition Using Score Distribution and Ranking
Hoai, M., & Zisserman, A. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-16814-2_1 · Cited by 86
Two streams Recurrent Neural Networks for Large-Scale Continuous Gesture Recognition
Chai, X., Liu, Z., Yin, F., Liu, Z., & Chen, X. (2016) · 2016 23rd International Conference on Pattern Recognition (ICPR) · doi:10.1109/icpr.2016.7899603 · Cited by 86 · cited 1×
“RGB-D-based motion recognition [8], [13], [19], [32], [33], [34], [35], [36], [37], thus, has attracted extensive attention thanks to the availability of additional depth cues.”
BackgroundA Unified Framework for Multi-Modal Isolated Gesture Recognition
Duan, J., Wan, J., Zhou, S., Guo, X., & Li, S. Z. (2018) · ACM Transactions on Multimedia Computing, Communications, and Applications · doi:10.1145/3131343 · Cited by 84 · cited 1×
“HUMAN motion recognition [1], [2], [3] is an attractive yet challenging task.”
BackgroundSelf-citationBilinear heterogeneous information machine for RGB-D action recognition
Kong, Y., & Fu, Y. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298708 · Cited by 80 · cited 1×
“[39] attempt to project the RGB-D data into a shared space to learn cross-modal features.”
BackgroundLarge-scale Isolated Gesture Recognition using pyramidal 3D convolutional networks
Zhu, G., Zhang, L., Mei, L., Shao, J., Song, J., & Shen, P. (2016) · 2016 23rd International Conference on Pattern Recognition (ICPR) · doi:10.1109/icpr.2016.7899601 · Cited by 71
Multi-Stream Deep Neural Networks for RGB-D Egocentric Action Recognition
Tang, Y., Wang, Z., Lu, J., Feng, J., & Zhou, J. (2019) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2018.2875441 · Cited by 71 · cited 1×
“We evaluate our method on four public RGB-D motion datasets: NTU RGB-D [56], NvGesture [57], Chalearn IsoGD [58], [59] and THU-READ [60] datasets.”
MethodA Transformer-Based Network for Dynamic Hand Gesture Recognition
D’Eusanio, A., Simoni, A., Pini, S., Borghi, G., Vezzani, R., & Cucchiara, R. (2020) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv50981.2020.00072 · Cited by 71
Large-Scale Multimodal Gesture Recognition Using Heterogeneous Networks
Wang, H., Wang, P., Song, Z., & Li, W. (2017) · 2017 IEEE International Conference on Computer Vision Workshops (ICCVW) · doi:10.1109/iccvw.2017.370 · Cited by 65 · cited 1×
“Recently, RGB-D-based motion recognition [4], [5], [6] has attracted much attention in computer vision due to the easy availability of depth data and its extensive application prospects such as video surveillance and human-object interfaces.”
BackgroundSelf-citationInfrared and 3D Skeleton Feature Fusion for RGB-D Action Recognition
De Boissiere, A. M., & Noumeir, R. (2020) · IEEE Access · doi:10.1109/access.2020.3023599 · Cited by 65 · cited 1×
“RGB-D-based motion recognition refers to jointly exploring the color and depth cues in multimodal data by constructing a multi-stream processing mechanism, thereby forming an informative joint representation for motion description [7], [8], [9].”
BackgroundLarge-Scale Gesture Recognition With a Fusion of RGB-D Data Based on Saliency Theory and C3D Model
Li, Y., Miao, Q., Tian, K., Fan, Y., Xu, X., Li, R., & Song, J. (2017) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2017.2749509 · Cited by 64 · cited 1×
“The NTU RGB-D is a large-scale human action dataset containing more than 56,000 multi-view RGB-D videos with 4 million frames that include 60 actions performed by 40 subjects.”
BackgroundJi, X., Cheng, J. S., Tao, D., Wu, X., & Feng, W. (2017) · Knowledge-Based Systems · doi:10.1016/j.knosys.2017.01.035 · Cited by 56
Decoupling and Recoupling Spatiotemporal Representation for RGB-D-based Motion Recognition
Zhou, B., Wang, P., Wan, J., Liang, Y., Wang, F., Zhang, D., Lei, Z., Li, H., & Jin, R. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.01952 · Cited by 53 · cited 9×
“Similar to most of the recent works [13], [19], [62], [63], we also evaluate the experimental results on the validation set for this dataset.”
BackgroundMethodResultInfluentialSelf-citationMaking Convolutional Networks Recurrent for Visual Sequence Learning
Yang, X., Molchanov, P. A., & Kautz, J. (2018) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2018.00677 · Cited by 51
One-shot learning gesture recognition from RGB-D data using bag of features
WanJun, RuanQiuqi, Liwei, L., & DengShuang (2013) · Journal of Machine Learning Research · doi:10.5555/2567709.2567743 · Cited by 49 · cited 1×
“Recently, RGB-D-based motion recognition [4], [5], [6] has attracted much attention in computer vision due to the easy availability of depth data and its extensive application prospects such as video surveillance and human-object interfaces.”
BackgroundLarge-scale gesture recognition with a fusion of RGB-D data based on optical flow and the C3D model
Li, Y., Miao, Q., Tian, K., Fan, Y., Xu, X., Ma, Z., & Song, J. (2017) · Pattern Recognition Letters · doi:10.1016/j.patrec.2017.12.003 · Cited by 48 · cited 1×
“Data augmentation, a data-driven and informed regularization strategy that artificially increases the number of training samples to improve the generalization ability of models, has been widely explored for better training of deep neural networks [20], [21], [22], [23].”
BackgroundChaLearn Looking at People: IsoGD and ConGD Large-Scale RGB-D Gesture Recognition
Wan, J., Lin, C., Wen, L., Li, Y., Miao, Q., Escalera, S., Anbarjafari, G., Guyon, I., Guo, G., & Li, S. Z. (2020) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2020.3012092 · Cited by 41
Self-citationRegional Attention with Architecture-Rebuilt 3D Network for RGB-D Gesture Recognition
Zhou, B., Li, Y., & Wan, J. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i4.16471 · Cited by 41
Self-citationA Hybrid Network for Large-Scale Action Recognition from RGB and Depth Modalities
Wang, H., Song, Z., Li, W., & Wang, P. (2020) · Sensors · doi:10.3390/s20113305 · Cited by 39 · cited 1×
“Similar to most of the recent works [13], [19], [61], [62], we also evaluate the experimental results on the validation set for this dataset.”
ResultSelf-citationThe Effects of Regularization and Data Augmentation are Class DependentPreprint
Balestriero, R., Bottou, L., & LeCun, Y. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2204.03632 · Cited by 39
Multi-Scale Attention 3D Convolutional Network for Multimodal Gesture Recognition
Chen, H., Li, Y., Fang, H., Xin, W., Lu, Z., & Miao, Q. (2022) · Sensors · doi:10.3390/s22062405 · Cited by 35
Hierarchically Decoupled Spatial-Temporal Contrast for Self-supervised Video Representation Learning
Zhang, Z., & Crandall, D. (2022) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv51458.2022.00105 · Cited by 29
Embedding Sequential Information into Spatiotemporal Features for Action Recognition
Ye, Y., & Tian, Y. (2016) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2016.142 · Cited by 24 · cited 1×
“[38] propose a Spatiotemporal-LSTM framework to learn motion patterns from temporal information with dense motion trajectories.”
BackgroundProgression Modelling for Online and Early Gesture Detection
Gupta, V., Dwivedi, S. K., Dabral, R., & Jain, A. (2019) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv.2019.00040 · Cited by 23
Learning Adaptive Hidden Layers for Mobile Gesture Recognition
Hu, T.-K., Lin, Y., & Hsiu, P.-C. (2018) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v32i1.12279 · Cited by 20 · cited 1×
“The late fusion strategies, including Addition and Multiplication, are simple yet effective multimodal fusion methods that are commonly used in RGB-D-based motion recognition [13], [52], [53].”
MethodDynamic video mix-up for cross-domain action recognition
Wu, H., Song, C., Yue, S., Wang, Z., Xiao, J., & Liu, Y. (2021) · Neurocomputing · doi:10.1016/j.neucom.2021.11.054 · Cited by 17 · cited 1×
“Wu et al. [31] mix two videos from different domains with fixed weights, to tackle the cross-domain problem in video.”
BackgroundAdaptive cross-fusion learning for multi-modal gesture recognition
Zhou, B., Wan, J., Liang, Y., & Guo, G. (2021) · Virtual Reality & Intelligent Hardware · doi:10.1016/j.vrih.2021.05.003 · Cited by 13 · cited 1×
“RGB-D-based motion recognition [8], [13], [19], [32], [33], [34], [35], [36], [37], [38], thus, has attracted extensive attention thanks to the availability of additional depth cues.”
BackgroundSelf-citationSpatiotemporal Joint Filter Decomposition in 3D Convolutional Neural Networks
Miao, Z., Wang, Z., Cheng, X., & Qiu, Q. (2021) · neural information processing systems · Cited by 4
KVT: k-NN Attention for Boosting Vision TransformersPreprint
Wang, P., Wang, X., Wang, F., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.00515 · Cited by 4 · cited 1×
“To reduce the redundant marginal information in captured temporal features, here we utilize a Trans-former structure based on Kvt [49].”
MethodSelf-citationHierarchically Decoupled Spatial-Temporal Contrast for Self-supervised Video Representation LearningPreprint
Zhang, Z., & Crandall, D. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.11261 · Cited by 3
Cited by
Gloss-free Sign Language Translation: Improving from Visual-Language Pretraining
Zhou, B., Chen, Z., Clapés, A., Wan, J., Liang, Y., Escalera, S., Lei, Z., & Zhang, D. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01908 · Cited by 130
“The goal of ISLR is to translate an isolated sign into a corresponding single sign language word [14, 15, 20, 22], which is somewhat similar to the isolated gesture recognition task [33, 18, 9, 38, 40, 39, 36].”
BackgroundSelf-citationRobust Multimodal Learning with Missing Modalities via Parameter-Efficient Adaptation
Reza, M. K., Prater-Bennette, A., & Asif, M. S. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3476487 · Cited by 78
Zhang, Y., & Wang, Y. (2025) · EURASIP Journal on Image and Video Processing · doi:10.1186/s13640-025-00677-0 · Cited by 26
Li, Y., Chen, H., Feng, G., & Miao, Q. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01917 · Cited by 24
“Although great progress has been made in this field, the attention has always been drawn to boost performance via better network structure [50, 48, 55, 56] or by introducing extra modalities of data [6, 18, 23, 19].”
BackgroundClassroom Behavior Detection Method Based on PLA-YOLO11n
Zhang, H., Zhou, G., He, W., & Deng, H. (2025) · Sensors · doi:10.3390/s25175386 · Cited by 14
“…multi-scale receptive fields; and (3) the integration of attention mechanisms to dynamically focus on key regions (e.g., hand-raising, head-down, interactive gestures) while suppressing background noise, thereby enhancing the accuracy and stability of action detection and classification [7–9].”
Metadata
- Authors
- Zhou, B., Wang, P., Wan, J., Liang, Y., & Wang, F.
- Affiliations
- Macau University of Science and Technology (MO)
- Amazon (United States) (US)
- Chinese Academy of Sciences (CN)
- Alibaba Group (United States) (US)
- Venue
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Published
- 2023-10
- Volume
- 45
- Issue
- 10
- Pages
- 11428-11442
- ISSN
- 0162-8828 (ISSN-L)
- 2160-9292
- 1939-3539
- OpenAlex
- W4376132550
- Cited by
- 38
- References
- 86
- Citations by year
- 2026: 7 · 2025: 19 · 2024: 9 · 2023: 3
- FWCI
- 3.45
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Pattern Analysis and Machine Intelligence (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Key Research and Development Plan (2021YFE0205700)
- External cooperation key project of Chinese Academy Sciences (173211KYSB20200002)
- Science and Technology Development Fund of Macau (0123/2022/A3, 0070/2020/AMJ, 0004/2020/A1)
- Guangdong Provincial Key R&D Programme (2019B010148001)
- Open Research Projects of Zhejiang Lab (2021KH0AB07)
- CCF-Zhipu AI Large Model OF (202219)
- Alibaba Group
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Human Pose and Action Recognition
- Physical Sciences › Computer Science › Human-Computer Interaction › Hand Gesture Recognition Systems
- Physical Sciences › Computer Science › Computer Graphics and Computer-Aided Design › Human Motion and Animation
- Keywords
- multimodal fusion, mixup, spatio-temporal representation learning, decision fusion
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.