Does Human Action Recognition Benefit from Pose Estimation?

By Angela Yao, Juergen Gall, Gabriele Fanelli, Luc Van Gool. 2011

200 citations recorded by the source corpus.

Subfields: Biomedical Engineering, 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 Biomedical Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Space-Time Representation of People Based on 3D Skeletal Data: A Review (2016, background): “real-world applications including Kinect-based gaming, as well as in computer vision research [22], [36].”
Transition Forests: Learning Discriminative Temporal Transitions for Action Recognition and Detection (2016, methodology): “To relieve this, [4, 33, 35] add a temporal regression term and maps appearance and pose features to vote in an action Hough space.”
A decision forest based feature selection framework for action recognition from RGB-depth cameras (2013, background): “The work in [12] has investigated these two categories of approaches using a wide range of features and has shown that even with high levels of noise, the recognition process benefits from using pose-based features.”
Posebits for Monocular Human Pose Estimation (2014, background): “Advantages of attributes have been demonstrated for object categorization [8, 25], and human action recognition [11, 28] with emphasis on transfer learning between classes.”
Instructing people for training gestural interactive systems (2012, unknown): “[33] but allows online gesture recognition.”
Fusing Geometric Features for Skeleton-Based Action Recognition Using Multilayer LSTM Networks (2018, unknown): “[7] that skeleton-based features outperform appearance-based features by using the same classifier on the same dataset.”
Learning Clip Representations for Skeleton-Based 3D Action Recognition (2018, methodology): “Previous works have also used pairs of joints to design relational features such as joint distance for interaction detection and recognition [41], [42].”
Action Recognition with Exemplar Based 2.5D Graph Matching (2012, background): “Inspired by [19], where it shows that the combined pose and appearance features help improve action recognition performance, our 2.”
G3Di: A Gaming Interaction Dataset with a Real Time Detection and Evaluation Framework (2014, background): “representation for human motion [3] [7] [10] [19] [20] so we focus on pose based approaches.”
Efficient Interaction Recognition through Positive Action Representation (2013, unknown): “To overcome this, [32] used relational body-pose features as introduced in [31], describing geometric relations between specific joints in a single pose or a short sequence of poses.”
Two-person interaction detection using body-pose features and multiple instance learning (2012, methodology): “[36] pointed out velocity features have the best accuracy for single activity recognition using clean motion capture data with reduced skeleton (=13 joints).”
3D skeleton-based human action classification: A survey (2016, background): “In this sense, “ one of the biggest challenges of using posed-based features is that semantically similar motions may not necessarily be numerically similar ” [61].”
On Geometric Features for Skeleton-Based Action Recognition Using Multilayer LSTM Networks (2017, unknown): “[34] extend [32]’s idea and modify pose-based features that are suitable for two persons’ interaction.”
First-Person Hand Action Benchmark with RGB-D Videos and 3D Hand Pose Annotations (2017, background): “In fullbody human action recognition it is known that using higher level and viewpoint invariant features such as body pose can benefit action recognition [54, 74, 79, 81, 86], although this has not yet been studied in detail for hands.”
Spatiotemporal representation of 3D skeleton joints-based action recognition using modified spherical harmonics (2016, background): “The studies recently conducted by (Alnowami et al., 2012; Obdrzalek et al., 2012; Yao et al., 2011) indicate that, compared with traditional 2D image data, the representation of skeleton joints only for action recognition provides better results.”
Pose-based human action recognition via sparse representation in dissimilarity space (2014, background): “in [32,33] raised the question of whether it is useful to perform pose estimation for the action recognition task or it is better to extract low-level appearance features directly from video data.”