Actions in context

By Marcin Marszalek, Ivan Laptev, Cordelia Schmid. 2009

1,409 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

Rescaling Egocentric Vision: Collection, Pipeline and Challenges for EPIC-KITCHENS-100 (2020, background): “4% Hollywood2 [60] 3 7 7 16 2,376 21.”
Dense-Captioning Events in Videos (2017, methodology): “1M 300s Thumos 15 [5] sports 21k 4s HMDB 51 [9] movie 7k 3s Hollywood 2 [10] movie 4k 20s MPII cooking [14] cooking 44 600s ActivityNet [1] human 20k 180s MPII MD [13] movie 68k 4s 68,375 X M-VAD [19] movie 49k 6s 55,904 X MSR-VTT [20] open 10k 20s 200,000 X MSVD [2] human 2k 10s 70,028 X YouCook [3”
HMDB: A large video database for human motion recognition (2011, background): “Computer vision could provide critical insight to this question as various approaches have been proposed that rely not just on motion cues like the two systems we have tested but also on single-frame shape-based cues, such as posture [18] and shape [19], and contextual information [13, 28].”
A dataset for Movie Description (2015, methodology): “Second, we propose to use semantic parsing of the sentence to extract training labels for recently proposed translation approach [56] for video description.”
AVA: A Video Dataset of Spatio-Temporally Localized Atomic Visual Actions (2017, background): “Most popular action clas-siﬁcation datasets, such as KTH [35], Weizmann [4], Hollywood-2 [26], HMDB [24], UCF101 [39] consist of short clips, manually trimmed to capture a single action.”
Beyond Gaussian Pyramid: Multi-skip Feature Stacking for action recognition (2014, methodology): “We use the standard splits with training and test videos provided by [17].”
Every Moment Counts: Dense Detailed Labeling of Actions in Complex Videos (2015, background): “The Hollywood (HOHA) [18], HMDB [14], UCF 101 [37], ActivityNet [5], and THUMOS [9] datasets exemplify this trend.”
The “Something Something” Video Database for Learning and Evaluating Visual Common Sense (2017, background): “For example, the task associated with the datasets described in [23, 12] is recognizing sports, and in [17] they include high-level, humancentered activities, such as “getting out of a car” or “fighting”.”
Weakly-Supervised Alignment of Video with Text (2015, methodology): “Textual data such as scripts, has been used for automatic video understanding, for example for action recognition [26, 31].”
Neural Motifs: Scene Graph Parsing with Global Context (2017, background): “Actions and relations have been a particularly fruitful source of context [27, 45], especially when combined with pose to create human-object interactions [43, 3].”
Large-Scale Weakly-Supervised Pre-Training for Video Action Recognition (2019, background): “[8, 21, 23, 47] analyzed movie scripts for automatic annotation of human actions for recognizing and localizing actions and actors.”
Watch and learn: Semi-supervised learning of object detectors from videos (2015, background): “One major body of work [7, 8, 12, 31, 46] tries to overcome this difﬁculty by leveraging the use of a set of pre-deﬁned attributes for image classes [8, 46] and additionally web-supervision and text [7, 12].”
Visual Semantic Role Labeling (2015, background): “Some of these study the task of full video action classiﬁcation [32, 14, 23, 26, 20, 31], while some [40, 18, 29] also study the task of detecting the agent doing the action.”
Weakly-Supervised Action Segmentation with Iterative Soft Boundary Assignment (2018, background): “This kind of labeling is much easier to obtain, and can even be automatically generated from other meta information [1, 11, 14].”
NeuralNetwork-Viterbi: A Framework for Weakly Supervised Video Learning (2018, background): “First attempts on weakly supervised learning have been made by [17, 21], who try to to obtain training examples based on movie scripts.”
Predictive-Corrective Networks for Action Detection (2017, background): “Predictive models: It has been shown that leveraging global contextual information can be used to improve image [26] or video [23] understanding.”
Modeling Actions through State Changes (2013, background): “[14] use scene context to improve action recognition performance.”