Modeling the Temporal Extent of Actions

By Scott Satkin, Martial Hebert. 2010

134 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Biomedical Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Video Representation Learning Using Discriminative Pooling (2018, background): “For example, Satkin and Hebert [35] explore the effect of temporal cropping of videos to regions of actions; however, assumes these regions are continuous.”
Weakly-Supervised Alignment of Video with Text (2015, background): “Finally, as mentioned in [13, 41], it is difficult to define exactly when an action occurs.”
Anticipating Human Activities Using Object Affordances for Reactive Robotic Response (2013, background): “It has been observed that temporal boundaries of actions are not precisely defined in practice, whether they are obtained automatically using weak-supervision [35] or by hand [36].”
Advanced Structured Prediction (2014, background): “Error at the boundaries does not necessarily indicate the flaw of our method as human labels are often imperfect (Satkin and Hebert, 2010).”
Leaving Some Stones Unturned: Dynamic Feature Prioritization for Activity Detection in Streaming Video (2016, background): “Common strategies are sliding temporal window search [24]–[26] or analyzing tracked objects [27]–[30].”
Human Action Recognition and Prediction: A Survey (2018, background): “As shown in [17], [222], not all frames are equally important for action recognition; only few of them are critical.”
Trespassing the Boundaries: Labeling Temporal Bounds for Object Interactions in Egocentric Video (2017, background): “Temporal Bounds in Non-Egocentric Datasets The leading work of Satkin and Hebert [24] ﬁrst pointed out that determining the temporal extent of an action is often subjective, and that action recognition results vary depending on the bounds used for training.”
Joint segmentation and classification of human actions in video (2011, background): “As a consequence, our results here are not directly comparable to some published numbers in the literature [12, 9, 20].”
Max-Margin Early Event Detectors (2012, background): “To the best of our knowledge, this is the first paper in the literature of computer vision that proposes a learning formulation for early event detection.”
Spatiotemporal Deformable Part Models for Action Detection (2013, background): “2As observed by [19], the correlation between action type and duration can cause researchers to overestimate the accuracy of action recognition when testing on temporally segmented video, since features inadvertently encode duration.”
High-level event recognition in unconstrained videos (2012, methodology): “Similar to the idea of sliding window search, Satkin and Hebert [120] located the best segment in a video for action training by exhaustively checking all possible segments of the video.”
The THUMOS challenge on action recognition for videos "in the wild" (2016, unknown): “While elimination of temporal clutter simpliﬁes the recognition problem, it becomes diﬃcult to predict the performance of different methods in real applications.”
Poselet Key-Framing: A Model for Human Activity Recognition (2013, background): “Key volumes: Recent video-based action recognition methods [12, 28] observed that limiting the bag of spatiotemporal interest point representation [9, 17] in a temporal segment boosts recognition performance.”
Learning latent spatio-temporal compositional model for human action recognition (2013, background): “Some other researchers learned temporal structures for action recognition [29, 35].”
Video Event Understanding Using Natural Language Descriptions (2013, background): “Weakly supervised action models Discriminative spatio temporal regions in videos or images to localize the actions in [27, 21, 28].”
DISCOVER: Discovering Important Segments for Classification of Video Events and Recounting (2014, methodology): “[17] uses a simple algorithm to evaluate the quality of a possible cropping using other uncropped training data, the goal is to filter irrelevant segments from training dataset.”
Learning latent temporal structure for complex event detection (2012, methodology): “Most similar to our method are recent works in video that learn discriminative models over temporal segments [12, 15, 16, 21].”
Temporal Localization of Actions with Actoms (2013, unknown): “[4, 6, 7, 10–12 , 38– 45”
UntrimmedNets for Weakly Supervised Action Recognition and Detection (2017, background): “More importantly, unlike object boundary, there is no precise deﬁnition about the exact temporal extent of actions [35, 36].”