Space-Variant Descriptor Sampling for Action Recognition Based on Saliency and Eye Movements

By E. Vig, Michael Dorr, David D. Cox. 2012

131 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

Training Object Class Detectors from Eye Tracking Data (2014, background): “Several authors have collected eye tracking data for video and shown that saliency maps computed from these data can improve action recognition [31,46].”
Action from Still Image Dataset and Inverse Optimal Control to Learn Task Specific Visual Scanpaths (2013, background): “Human gaze pattern annotations have been collected for both static images[11, 13, 14, 12, 26, 18] and for video[19, 23, 15], see [24] for a recent overview.”
Human Action Recognition in Unconstrained Videos by Explicit Motion Modeling (2015, background): “A few others focused on the selection or sampling of more effective local volumes [17], [18] or higher-level attribute representations [19], [20].”
What are the Visual Features Underlying Human Versus Machine Vision? (2017, background): “Researchers have found that in cases like these human perceptual judgments can augment the performance of vision models [35], resulting in significant gains in face recognition and localization [24, 2, 14], action recognition [31], object detection and segmentation [20, 26, 32].”
Recurrent Mixture Density Network for Spatiotemporal Visual Attention (2016, background): “The potential applications of automatic saliency map prediction from videos are many.”
Perceptual Annotation: Measuring Human Vision to Improve Computer Vision (2014, background): “, eye movements used to define discriminative image regions for feature extraction [14], [15], structured domain knowledge from human experts [16], [17], [18], and models of the typical human annotation process itself [19].”
LEAR-INRIA submission for the THUMOS workshop (2013, result): “This results in similar effects as sampling features based on visual saliency maps [6, 12].”
Human Action Recognition across Datasets by Foreground-Weighted Histogram Decomposition (2014, methodology): “[26] used saliency to remove features from nonsalient regions, however, their method requires thresholding of saliency maps while each of the remaining features contribute equally to the final representation.”
A Robust and Efficient Video Representation for Action Recognition (2015, unknown): “This gives us similar effects as sampling features based on visual saliency maps (Mathe and Sminchisescu, 2012; Vig et al, 2012).”
Attentive Systems: A Survey (2017, methodology): “Likewise, Vig et al. (2012) employ saliency-mapping algorithms to find informative regions and descriptors corresponding to these regions are either used exclusively, or are given greater representational weights with additional codebook vectors.”
Eye tracking assisted extraction of attentionally important objects from videos (2015, background): “In a free viewing task, eye movements are biased towards high level semantics [22, 29] in static and dynamic scenes.”
Activity representation with motion hierarchies (2013, unknown): “Our method is outperformed on the Hollywood 2 dataset by [Vig et al, 2012], who use BOF with saliency masks, [Jiang et al, 2012a], who combine multiple features, and [Wang et al, 2013], who also combine multiple trajectory features and uses spatio-temporal pyramids.”
Self-Paced Learning with Diversity (2014, unknown): “2012 [25] 59.”
Dynamic Eye Movement Datasets and Learnt Saliency Models for Visual Action Recognition (2012, background): “See [17] for alternative work, published in this conference, on eye movement data collection and saliency models for action recognition.”