Video Action Detection with Relational Dynamic-Poselets

By Limin Wang, Yu Qiao, Xiaoou Tang. 2014

138 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Control and Systems Engineering

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

In Control and Systems Engineering

Citations in the corpus

Dense-Captioning Events in Videos (2017, background): “With the introduction of large scale activity datasets [22, 19, 12, 4], it has become possible to categorize videos into a discrete set of action categories [27, 10, 9, 40, 35].”
Temporal Segment Networks: Towards Good Practices for Deep Action Recognition (2016, background): “[30] designed a sequential skeleton model (SSM) to capture the relations among dynamic-poselets, and performed spatio-temporal action detection.”
Finding action tubes (2014, methodology): “[42] propose a unified approach to discover effective action parts using dynamical poselets and model their relations.”
Towards Weakly-Supervised Action Localization (2016, background): “Recently, significant progress was achieved, see for example [5,6,7].”
Temporal Segment Networks for Action Recognition in Videos (2017, background): “temporal structure of video for action recognition [19], [20], [21], [22], [59], [60].”
Learning to Track for Spatio-Temporal Action Localization (2015, neutral): “[45] first use a temporal sliding window and then model the relations between dynamic-poselets.”
Analysing Domain Shift Factors between Videos and Images for Object Detection (2015, unknown): “However, learning a detector with weak supervision is very challenging and current performance is still well below fully supervised methods [7], [38], [45].”
Temporal Localization of Fine-Grained Actions in Videos by Domain Transfer from Web Images (2015, background): “Wang et al. [37] used dynamic poselets which took motion and pose into account.”
Deep Temporal Linear Encoding Networks (2016, methodology): “To overcome such issues, several techniques have been proposed to model the temporal structure for action recognition, such as the actom sequence model [10] which considers sequence of histograms; temporal action decomposition [21] which exploits the temporal structure of human actions by temporally”
Spot On: Action Localization from Pointly-Supervised Proposals (2016, background): “Independent training data can be brought in to automatically detect individual persons for action localization [3,17,18].”
Learning Activity Progression in LSTMs for Activity Detection and Early Detection (2016, methodology): “, encoding video clips using a bagof-words representation over local space-time features and training SVM classifiers [11, 13, 23], or modeling human actions as space-time structures [1, 10, 12, 17, 24, 25].”
Actionness Estimation Using Hybrid Fully Convolutional Networks (2016, methodology): “[39] proposed a unified approach to perform action detection and pose estimation by using dynamic-poselets and modeling their relations.”
Holistic Large Scale Video Understanding (2019, methodology): “Several other techniques were proposed to model the temporal structure in an efﬁcient way, such as the actom sequence model [15]; temporal action decomposition [31]; dynamic poselets [46]; ranking machines [14].”
R-C3D: Region Convolutional 3D Network for Temporal Activity Detection (2017, background): “…approaches suffer from one or more of the following major drawbacks: they do not learn deep representations in an end-to-end fashion, but rather use hand-crafted features [33, 34], or deep features like VGG [28], ResNet [8], C3D [32] etc. , learned separately on image/video classiﬁcation tasks.”
Multi-region Two-Stream R-CNN for Action Detection (2016, methodology): “[7] apply dynamic poselets and a sequential skeleton model to jointly detect actions and poses.”