Multi-view Super Vector for Action Recognition

By Zhuowei Cai, Limin Wang, Xiaojiang Peng, Yu Qiao. 2014

202 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. G. Wang
3. Xiaogang Wang
4. Dacheng Tao
5. Yi Yang
6. Yu Qiao
7. Ming-Hsuan Yang
8. Cordelia Schmid

Citations in the corpus

• Deep Multimodal Feature Analysis for Action Recognition in RGB+D Videos (2016, methodology): “In this respect, we can enhance the representation by incorporating the modality-specific components of respective modalities [6, 51].”
• Action recognition with trajectory-pooled deep-convolutional descriptors (2015, neutral): “…flow HMDB51 UCF101 STIP+BoVW [15] 23.0% STIP+BoVW [26] 43.9% Motionlets [35] 42.1% Deep Net [12] 63.3% DT+BoVW [30] 46.6% DT+VLAD [3] 79.9% DT+MVSV [3] 55.9% DT+MVSV [3] 83.5% iDT+FV [31] 57.2% iDT+FV [32] 85.9% iDT+HSV [21] 61.1% iDT+HSV [21] 87.9% Two Stream [24] 59.4% Two Stream [24] 88.0%…”
• Bag of visual words and fusion methods for action recognition: Comprehensive study and good practice (2014, neutral): “We randomly sample 50 videos from the UCF101 dataset and report the total time for these videos.”
• Temporal Segment Networks for Action Recognition in Videos (2017, methodology): “DT+MVSV [46] 55:9% DT+MVSV [46] 83:5% object+motion [73] 71.”
• ActionVLAD: Learning Spatio-Temporal Aggregation for Action Classification (2017, unknown): “…63.2 RNN+FV [29] 88.0 54.3 Transformations [59] 92.4 62.0 LTC [51] 91.7 64.8 KVMF [63] 93.1 63.3 ActionVLAD (LateFuse, VGG-16) 92.7 66.9 DT+MVSV [9] 83.5 55.9 iDT+FV [53] 85.9 57.2 iDT+HSV [33] 87.9 61.1 MoFAP [56] 88.3 61.7 C3D+iDT [49] 90.4 - Two-Stream Fusion+iDT [15] 93.5 69.2 LTC+iDT [51]…”
• Human Action Recognition in Unconstrained Videos by Explicit Motion Modeling (2015, methodology): “Compared with recent works on fusing multiple super vectors [58] and improved VLAD encoding [60], we also achieve better result with clear margins.”
• Deep Temporal Linear Encoding Networks (2016, unknown): “It is also interesting to see that, TLE:Bilinear improves the C3D ConvNet performance over Method UCF101 HMDB51 DT+MVSM [2] 83.5 55.9 iDT+FV [34] 85.9 57.2 Two Stream [24] 88.0 59.4 VideoDarwin [8] − 63.7 C3D [32] 82.3 56.8 Two Stream+LSTM [40] 88.6 − F ST CV (SCI fusion) [29] 88.1 59.1 TDD+FV [36]…”
• What do 15,000 object categories tell us about classifying and localizing actions? (2015, unknown): “Three of the best performers on UCF101 [1, 32, 55] propose improved encoding methods.”
• DynamoNet: Dynamic Action and Motion Network (2019, unknown): “Method UCF101 HMDB51 DT+MVSM [2] 83.5 55.9 iDT+FV [48] 85.9 57.2 C3D [42] 82.3 56.8 C3D+iDT [42] 90.4 − LTC+iDT [42] 92.4 67.2 Conv Fusion [11] 82.6 56.8 Two Stream [36] 88.6 − TDD+FV [49] 90.3 63.2 RGB+Flow-TSN [50] 94.0 68.5 ST-ResNet [10] 93.5 66.4 TSN [50] 94.2 69.5 RGB-I3D [3] 95.6 74.8…”
• Learning Long-Term Dependencies for Action Recognition with a Biologically-Inspired Deep Network (2016, unknown): “9% DT+MVSV [4] 55 .”
• Jointly Learning Heterogeneous Features for RGB-D Activity Recognition (2015, methodology): “Although the CCA in [3] is mostly close to ours, it is not for discriminative learning, and moreover it assumes that the specific component for each fea-”
• Temporal Segment Networks: Towards Good Practices for Deep Action Recognition (2016, unknown): “HMDB51 UCF101 DT+MVSV [37] 55 .”
• Spatio-Temporal Channel Correlation Networks for Action Classification (2018, unknown): “UCF101 HMDB51 DT+MVSM [34] 83.5 55.9 iDT+FV [35] 85.9 57.2 C3D [2] 82.3 56.8 Conv Fusion [17] 82.6 56.8 Two Stream [16] 88.6 − TDD+FV [36] 90.3 63.2 TSN-RGB [37] 85.7 − P3D [38] 88. fully connected layers (C3D w/ BN), and RGB-I3D which is without pretraining on the ImageNet (RGB-I3D w/o ImageNet)…”