Retrieving actions in movies

By Ivan Laptev, Patrick Pérez. IEEE International Conference on Computer Vision, 2007

510 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/c6cd1a1c1e6bde8e06120c431d4e816e4d67f249

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 Computer Vision and Pattern Recognition

Citations in the corpus

Learning the semantics of object–action relations by observation (2011, background): “…much attention in recent years and many contributions exist, but are often unrelated to manipulation recognition (Calinon and Billard 2004, 2005; Hakeem and Shah 2005; Maurer et al. 2005; Calinon and Billard 2007; Laptev and Perez 2007; Niebles et al. 2008; Dee et al. 2009; Gilbert et al. 2011).”
Human Action Localization with Sparse Spatial Supervision (2017, background): “Initial attempts for temporal and spatio-temporal action localization are based on a sliding-window scheme and focus on improving the search complexity [23], [24], [25], [26].”
YouTube2Text: Recognizing and Describing Arbitrary Activities Using Semantic Hierarchies and Zero-Shot Recognition (2013, methodology): “Previously published activity recognition methods that work on datasets such as KTH [26], Drinking and Smoking [16] and UCF50 [24] have a very limited recognition vocabulary of activity classes, ranging from 6 to 12 action classes.”
A Spatio-Temporal Descriptor Based on 3D-Gradients (2008, unknown): “In later work, Laptev et al. [10, 11] applied a coarse quantization to gradient orientations.”
High Five: Recognising human interactions in TV shows (2010, methodology): “Our experiments show that using structured learning improves the retrieval results compared to using the interaction classifiers independently.”
A latent model of discriminative aspect (2009, background): “There are two important threads: first, one might recognize activities using discriminative methods applied to image features either computed by segmenting the body (for example [1]), or from characteristic motion fields (for example [19]).”
LabelMe: Online Image Annotation and Applications (2010, background): “Another example is the database in [31], and later extended in [30], which was collected from Hollywood movies.”
AMTnet: Action-Micro-Tube Regression by End-to-end Trainable Deep Architecture (2017, background): “While many works concern either spatial action localisation [18, 32, 10, 25] in trimmed videos or temporal localisation [17, 3, 27, 19, 31, 23, 35, 34] in untrimmed videos, only a handful number of methods have been proposed to tackle both problems jointly.”
A Data-Driven Approach for Event Prediction (2010, background): “Several works have proposed models to study, characterize, and classify human actions ranging from constrained environments [3, 4] to actions in the wild such as TV shows, sporting events, and cluttered backgrounds [5, 6].”
Human Action Recognition and Prediction: A Survey (2018, methodology): “using the so-called bag-of-words model, which captures local motion patterns using a histogram of visual words [14,137,227,135,170].”
Human Activity Recognition with Metric Learning (2008, background): “Such appearance features include: braids [8]; characteristic spatio-temporal volumes [9]; motion energy images [10]; motion history images [10]; spatio-temporal interest points [11,12,13]; nonlinear dimensionality reduced stacks of silhouettes [14]; an extended radon transform [15]; and silhouette h”
Subtitle-free Movie to Script Alignment (2009, background): “We have presented a framework that can be extended as visual object recognition improves [8] or recognition of actions/interactions improves [12, 13], so that more correspondences can be found between nouns/verbs in the script and visual detections.”
Weakly Supervised Learning of Interactions between Humans and Objects (2012, unknown): “Most existing methods for action recognition either learn a spatio-temporal model of an action [3], [4], [5] or are based on human pose [6], [7].”
Generating Natural-Language Video Descriptions Using Text-Mined Knowledge (2013, unknown): “All the published activity recognition methods that work on datasets such as KTH (Schuldt et al., 2004), Drinking and Smoking (Laptev and Perez, 2007) and UCF50 (Reddy and Shah, 2012) have a very limited recognition vocabulary of activity classes.”
Efﬁcient feature extraction, encoding and classiﬁcation for action recognition (unknown): “Recent methods show signiﬁcant progress towards action recognition in realistic and challenging videos from YouTube, movies and TV [16, 17, 8, 22, 29, 31, 35].”
A Robust and Efficient Video Representation for Action Recognition (2015, unknown): “…dataset for action localization is extracted from the movie Coffee and Cigarettes , and contains annotations for the actions drinking and smoking (Laptev and P´erez, 2007 To measure performance we compute the average precision (AP) score as in (Duchenne et al, 2009; Gaidon et al, 2011; Kl ¨…”
Efficient Feature Extraction, Encoding, and Classification for Action Recognition (2014, background): “Striking indications of this development include 6000 years of video uploaded to YouTube yearly1 and millions of surveillance cameras installed only in the UK.”