Efficient visual event detection using volumetric features

By Yan Ke, Rahul Sukthankar, Martial Hebert. Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1, 2005

643 citations recorded by the source corpus.

This paper studies the use of volumetric features as an alternative to popular local descriptor approaches for event detection in video sequences. Motivated by the recent success of similar ideas in object detection on static images, we generalize the notion of 2D box features to 3D spatio-temporal volumetric features. This general framework enables us to do real-time video analysis. We construct a realtime event detector for each action of interest by learning a cascade of filters based on volumetric features that efficiently scans video sequences in space and time. This event detector recognizes actions that are traditionally problematic for interest point methods - such as smooth motions where insufficient space-time interest points are available. Our experiments demonstrate that the technique accurately detects actions on real-world sequences and is robust to changes in viewpoint, scale and action speed. We also adapt our technique to the related task of human action classification and confirm that it achieves performance comparable to a current interest point based human activity recognizer on a standard database of human activities.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/27f2c4969e424242200eb1901143aeda7d0a4030

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

Recent advances and trends in visual tracking: A review (2011, background): “Ke et al. [45] studied the use of volumetric features for event detection in video sequences.”
Action recognition by learning mid-level motion features (2008, neutral): “[11] Splits 62.”
Discovering Motion Primitives for Unsupervised Grouping and One-Shot Learning of Human Actions, Gestures, and Expressions (2013, neutral): “Other examples of work in this category include Ke et al. [16], who proposed the learning of a cascade of filters using space-time volumetric features, effectively performing action detection as well as precise spatiotemporal localization.”
TORNADO: A Spatio-Temporal Convolutional Regression Network for Video Action Proposal (2017, background): “Video based Action Proposal: Early success in action detection is based on exhaustive search using sliding cuboids [13, 15, 35].”
Actions as Space-Time Shapes (2007, neutral): “be advantageous over many existing methods ([3], [7], [10], [11], [15], [ 16 ], [17], [21], [20], [24], [25],”
MoSIFT: Recognizing Human Actions in Surveillance Videos (2009, methodology): “The spatio-temporal features share some of the same desirable properties as local features of images and also achieve good performance on human action recognition tasks [1, 2, 3, 4, 5, 6, 7].”
Recognising action as clouds of space-time interest points (2009, background): “Spatio-temporal shape template based approaches essentially treat the action recognition problem as a 3-D object recognition problem by representing action using features extracted from spatio-temporal volume of an action sequence [7, 8, 21].”
Human Action Segmentation and Recognition Using Discriminative Semi-Markov Models (2011, neutral): “[10] ICCV’05 cascade classifier, spatio-temporal volumetric features, feature selection 0.”
A Scalable Approach to Activity Recognition based on Object Use (2007, background): “A common theme in these works is the exploration of spatio-temporal video features [3, 5, 11, 21, 25].”
Spatiotemporal Deformable Part Models for Action Detection (2013, background): “This paper focuses on the related problem of action detection [7, 21], sometimes termed action localization [12] or event detection [8, 9], where the goal is to detect every occurrence of a given action within a long video, and to localize each detection both in space and time.”
Retrieving actions in movies (2007, methodology): “The method in [19], however, provides action localization in time only while in [10] action detection was studied for simple scenes with restricted view variations.”
A flexible model for training action localization with varying levels of supervision (2018, methodology): “Initial attempts [20, 22] scanned the video clip with a 3D sliding window detector on top of volumetric features.”
A Unified Framework for Gesture Recognition and Spatiotemporal Gesture Segmentation (2009, methodology): “Ke et al. [24] propose modeling actions as rigid 3D patterns and extracting action features using 3D extensions of the well-known 2D rectangle filters [25].”
SIFT-Bag kernel for video event analysis (2008, background): “The experiments on the TRECVID 2005 corpus demonstrate that the mean average precision is boosted from the best reported 38.2% in [36] to 60.4% based on our new framework.”
Canonical Correlation Analysis of Video Volume Tensors for Action Categorization and Detection (2009, positive): “applied the spatiotemporal volumetric features [17].”
A 3-dimensional sift descriptor and its application to action recognition (2007, methodology): “In the past, solutions to the action recognition problem have utilized features such as optical flow [4], silhouette shapes [12], volume based representation [5], etc.”