Evolutionary joint selection to improve human action recognition with RGB-D devices
By Alexandros André Chaaraoui, José Ramón Padilla-López, Pau Climent-Pérez, Francisco Flórez-Revuelta. 2014
153 citations recorded by the source corpus.
Interest in RGB-D devices is increasing due to their low price and the wide range of possible applications that come along. These devices provide a marker-less body pose estimation by means of skeletal data consisting of 3D positions of body joints. These can be further used for pose, gesture or action recognition. In this work, an evolutionary algorithm is used to determine the optimal subset of skeleton joints, taking into account the topological structure of the skeleton, in order to improve the final success rate. The proposed method has been validated using a state-of-the-art RGB action recognition approach, and applying it to the MSR-Action3D dataset. Results show that the proposed algorithm is able to significantly improve the initial recognition rate and to yield similar or better success rates than the state-of-the-art methods.
Subfields: Computer Vision and Pattern Recognition, Human-Computer Interaction
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 Human-Computer Interaction
Citations in the corpus
- Space-Time Representation of People Based on 3D Skeletal Data: A Review (2016, methodology): “[165] Evolutionary Joint Selection BoW Lowlv Dict X”
- A Human Activity Recognition System Using Skeleton Data from RGBD Sensors (2016, methodology): “[15, 36] exploits more complex algorithms, considering Table 9: Three subsets of gestures fromMSR Action3D dataset.”
- Human Action Recognition From Various Data Modalities: A Review (2020, background): “(a)-(c) are originally shown in [181], [182], [183].”
- Mining Mid-Level Features for Action Recognition Based on Effective Skeleton Representation (2014, methodology): “[18] proposed to use an evolutionary algorithm to determine the optimal subset of skeleton joints, taking into account the topological structure of the skeleton.”
- Action2Motion: Conditioned Generation of 3D Human Motions (2020, methodology): “In [32], pair-wise relative positions of joints are used to represent human skeleton, while in [4], uninformative keypoints are further pruned during modeling.”
- A discussion on the validation tests employed to compare human action recognition methods using the MSR Action3D dataset (2014, background): “7.8 87.1 79.9 - Table7.Leave-one-actor-out cross-validation test Method Year AS1 AS2 AS3 Avg. All Evolutionary joint selection to improve human action recognition with RGB-D devices, Chaaraoui et al. [67] 2014 91.46 91.78 97.13 93.46 - Fusion of Skeletal and Silhouette-Based Features for Human Actio”
- Clustered Spatio-temporal Manifolds for Online Action Recognition (2014, methodology): “Key poses [7], [8] are a representative subset of the template poses selected by clustering techniques.”
- Real-Time Multi-scale Action Detection from 3D Skeleton Data (2015, methodology): “Wrapper methods [13, 6] rely on a learning algorithm to decide a subset of important features.”
- Simultaneous Feature and Body-Part Learning for real-time robot awareness of human behaviors (2017, methodology): “Several methods were also proposed to select more descriptive human body joints [2], [10], [11], [20], [21], [22], [23].”
- Low-Rank Tensor Subspace Learning for RGB-D Action Recognition (2016, background): “Along with the cost-efficient by Kinect sensor, the RGB-D camera which captures both RGB and depth data has been employed widely for human action recognition [5]–[8].”
- A Vision-Based System for Intelligent Monitoring: Human Behaviour Analysis and Privacy by Context (2014, unknown): “In fact, in previous work, we have already obtained successful results for gesture recognition relying on skeletal information and single-view setups [49].”