Activity representation with motion hierarchies

By Adrien Gaidon, Zaïd Harchaoui, Cordelia Schmid. 2013

120 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

1. Bernard Ghanem
2. Jun Liu
3. Juan Carlos Niebles
4. Li Fei-Fei
5. Qiuhong Ke
6. Silvio Savarese
7. Hossein Rahmani
8. Victor Escorcia

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Ming-Hsuan Yang
3. Li Fei-Fei
4. Jitendra Malik
5. Dacheng Tao
6. Andrew Zisserman
7. Cordelia Schmid
8. Xiaogang Wang

Citations in the corpus

• Beyond Gaussian Pyramid: Multi-skip Feature Stacking for action recognition (2014, neutral): “[6] 85.”
• Temporal Relational Reasoning in Videos (2017, neutral): “Besides, many previous works model the temporal structures of videos for action recognition and detection using bag of words, motion atoms, or action grammar [5, 16, 26, 6, 27].”
• A Review of Human Activity Recognition Methods (2015, methodology): “An unsupervised method for learning human activities from short tracklets was proposed by Gaidon et al. (2014).”
• A Robust and Efficient Video Representation for Action Recognition (2015, unknown): “Gaidon et al (2013) use efficient image-stitching techniques to compute the approximate motion of the background plane and generate stabilized videos before extracting dense trajectories (Wang et al, 2013a) for activity recognition.”
• Action Recognition by Hierarchical Mid-Level Action Elements (2015, methodology): “Among all of the methods, [32], [24], [11] and our full model use hierarchical structures.”
• AdaScan: Adaptive Scan Pooling in Deep Convolutional Neural Networks for Human Action Recognition in Videos (2016, methodology): “Such solutions either use Latent Variable Models [22, 36, 9, 30, 19], which require an additional inference step during learning, or employ a variant of Recurrent Neural Networks (RNN) [29, 49] which have intermediate hidden states that are not immediately inter-pretable.”
• Going deeper into action recognition: A survey (2016, unknown): “The motion an object and its location is handled separately through the Dosaral Stream Goodale and Milner (2003). A class of deep neural networks opted for separating appearance based information from motion related ones for action recognition Simonyan and Zisserman (2014). Simonyan and Zisserman (2”
• A Comprehensive Survey of Vision-Based Human Action Recognition Methods (2019, methodology): “[62] used split clustering to analyze local motion trajectories, and used the clustering results to represent different motion levels in calculating human action features.”
• Sympathy for the Details: Dense Trajectories and Hybrid Classification Architectures for Action Recognition (2016, methodology): “As in [67,68,5,6], we report mAP for the positive classes (mAP+) using the train/test split provided by the dataset authors.”
• Human Action Recognition From Various Data Modalities: A Review (2020, methodology): “Researchers have also made efforts in further improving the accuracy and effectiveness of the iDT algorithm [91], [92], [93].”
• SST: Single-Stream Temporal Action Proposals (2017, background): “Many approaches use a global representation [33], but some try to model the temporal structure of motions to achieve classification [10].”