Cross-dataset action detection

By Liangliang Cao, Zicheng Liu, Thomas S. Huang. 2010

227 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

In Computer Vision and Pattern Recognition

Citations in the corpus

RGB-D-based action recognition datasets: A survey (2016, methodology): “The cross-dataset scheme has already been adopted on some algorithms for action recognition in 2D videos [110,111], however, to our best knowledge, there is no report of its usage on RGB-D video datasets.”
RGB-D-based Human Motion Recognition with Deep Learning: A Survey (2017, background): “Even though some works have studied this topic (Cao et al., 2010; Sultani and Saleemi, 2014; Zhang et al., 2017a), they are all based on hand-crafted features, and the results are far from satisfactory due to the large distribution variances between different datasets, including different scenarios,”
Towards Weakly-Supervised Action Localization (2016, background): “Initial attempts for temporal and spatio-temporal action localization are based on a sliding-window scheme and focus on improving the search complexity [11,18,19,20].”
Action Tubelet Detector for Spatio-Temporal Action Localization (2017, unknown): “Initial approaches for spatio-temporal action localization are extensions of the sliding window scheme [2, 16], requiring strong assumptions such as a cuboid shape, i.”
Reshaping Visual Datasets for Domain Adaptation (2013, background): “One pitfall of this ad hoc practice is that a dataset could be an agglomeration of several distinctive domains.”
Multi-Modal Domain Adaptation for Fine-Grained Action Recognition (2020, background): “Before deep-learning, UDA for action recognition used shallow models to align source and target distributions of handcrafted features [4, 11, 64].”
What do 15,000 object categories tell us about classifying and localizing actions? (2015, unknown): “Though most of the emphasis in the action literature has been on classification, progress has also been made in action localization [2, 15, 22, 29, 44].”
Online Action Detection (2016, unknown): “MSRII [22], for example, contains only 54 short video sequences with only three action classes.”
Learning to Share Latent Tasks for Action Recognition (2013, background): “Human action recognition is an important problem in computer vision and numerous methods have been proposed to tackle it [13, 3, 17, 31, 28, 14, 25, 30].”
Transfer Learning for Activity Recognition: A Survey (2013, methodology): “Applying transfer learning to video clips from different spaces usually results in handling issues of background differences [9, 74, 75] and/or issues of camera view angle [37].”
Action Localization with Tubelets from Motion (2014, methodology): “Precision/recall: Comparison with [4, 24] for the 3 classes on MSR-II.”
Learning Tags from Unsegmented Videos of Multiple Human Actions (2011, background): “pre-segmented or strongly labeled data for training [30], [18], [29], [7].”
Bayesian Joint Modelling for Object Localisation in Weakly Labelled Images (2015, methodology): “In contrast to these discriminative approaches, we exploit a simple and efficient Bayesian adaptation approach similar in spirit to [34], [40].”
Watching Unlabeled Video Helps Learn New Human Actions from Very Few Labeled Snapshots (2013, background): “Domain adaptation can improve event learning across two domains of videos, such as Web videos to consumer videos [10] or one benchmark dataset to another [5], and it can also help train an object detector from videos [21].”
Exploring the Trade-off Between Accuracy and Observational Latency in Action Recognition (2012, methodology): “Cheema et al. (2011) propose to learn weights for contour-based distinctive key poses and classify using a weighted voting system. Lv and Nevatia (2007) represent actions using a series of 2D human poses and perform silhouette matching between input and key frames.”
Measuring and reducing observational latency when recognizing actions (2011, background): “Evaluating data on video where the action is being performed continuously eliminates the ambiguity that occurs as the user transitions into different actions.”
Harnessing Lab Knowledge for Real-World Action Recognition (2014, background): “Another work by Cao et al. (2005) is focused on using a lab dataset for action detection on a real-world dataset.”
Real-Time Gesture Recognition from Depth Data through Key Poses Learning and Decision Forests (2012, methodology): “This approach has attracted a lot of attention in the past few years [5], [6], [7], [8] and an example of largely used local feature is the SpaceTime Interest Point (STIP) presented by Laptev and Lindeberg [9].”