Motionlets: Mid-level 3D Parts for Human Motion Recognition

By Limin Wang, Yu Qiao, Xiaoou Tang. 2013

162 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

Citations in the corpus

Towards Good Practices for Very Deep Two-Stream ConvNets (2015, background): “The past several years have witnessed great progress on action recognition from short clips [8, 9, 12, 16, 17, 18, 19].”
Action recognition with trajectory-pooled deep-convolutional descriptors (2015, background): “To overcome this issue, several mid-level and high-level video representations have been proposed such as Action Bank [22], Dynamic-Poselets [37], Motionlets [35], Motion Atoms and Phrases [34], and Actons [42].”
Bag of visual words and fusion methods for action recognition: Comprehensive study and good practice (2014, neutral): “[52] 2013 42.”
Temporal Segment Networks for Action Recognition in Videos (2017, background): “mid-level representations have been proposed for action recognition [3], [23], [47], [48], [49], [50], [51].”
Real-Time Action Recognition with Enhanced Motion Vector CNNs (2016, background): “[34, 30, 31] exploited mid-level representations by proposing MoFAP and Motionlets.”
Hierarchical Clustering Multi-Task Learning for Joint Human Action Grouping and Recognition (2017, unknown): “9 Wang [52] 42.”
Learning to Share Latent Tasks for Action Recognition (2013, background): “Human action recognition is an important problem in computer vision and numerous methods have been proposed to tackle it [13, 3, 17, 31, 28, 14, 25, 30].”
Simple to Complex Transfer Learning for Action Recognition (2016, unknown): “[30] used spatiotemporal parts for recognizing human actions.”
Learning Expressionlets on Spatio-temporal Manifold for Dynamic Facial Expression Recognition (2014, methodology): “We also conduct comparisons with four popular local spatio-temporal descriptors: LBP-TOP [34], 3D SIFT [23], HOG 3D [14], HOE [27].”
Actionness Estimation Using Hybrid Fully Convolutional Networks (2016, background): “Most of the research works focused on the problem of action classification [7, 9, 22, 37, 38, 41], which aims at predicting an action label given a video.”
Listen to Look: Action Recognition by Previewing Audio (2019, methodology): “Research has transitioned from initial methods using hand-crafted local spatiotemporal features [35, 66, 61] to mid-level descriptors [46, 29, 62], and more recently to deep video representations learned end-to-end [50, 30, 16].”
Action Recognition with Actons (2013, background): “, discriminative parts/patches [22, 23, 29, 28] or semantical attributes [25, 15, 14]) for image/video recognition has been a recent active area in computer vision.”
Latent Hierarchical Model of Temporal Structure for Complex Activity Classification (2014, methodology): “In addition to these low-level local features and BoVW representation, there were some research works on mid-level and high-level representations such as motionlet [32], motion atom and phrase [15], action bank [33], and so on.”
Activity Auto-Completion: Predicting Human Activities from Partial Videos (2015, background): “Recently, some works focus on extracting mid-level features [5, 21, 10, 24, 11, 2].”
UntrimmedNets for Weakly Supervised Action Recognition and Detection (2017, background): “Action recognition in videos has attracted a lot of research attention in the past few years, and much progress has been made in computer vision community, on both aspects of hand-crafted representations [25, 43, 44, 46] and deeply-learned representations [21, 38, 40, 48].”
Dense Procedure Captioning in Narrated Instructional Videos (2019, background): “Previous works on video understanding tend to recognize actions in video clips by detecting pose (Wang et al., 2013a; Packer et al., 2012) and motion (Wang et al., 2013b; Yang et al., 2013) or both (Wang et al., 2014) and ﬁne-grained features(Rohrbach et al., 2016).”
Learning Long-Term Dependencies for Action Recognition with a Biologically-Inspired Deep Network (2016, unknown): “2.29% 92.55% zMerge the spatial, temporal and warp temporal streams [49]. Table 6. Comparison of sDTD to the state-of-the-art methods. HMDB51 UCF101 STIP+BoF [20] 23:0% STIP+BoF [20] 43:9% Motionlets [45] 42:1% Deep Net [18] 63:3% DT+BoF [42] 46:6% DT+VLAD [3] 79:9% DT+MVSV [3] 55:9% DT+MVSV [3] 83:”