Abstract:ABSTRACTEvent detection plays an essential role in video content analysis and remains a challenging open problem. In particular, the study on detecting human-related video events in complex scenes with both a crowd of people and dynamic motion is still limited. In this paper, we investigate detecting video events that involve elementary human actions, e.g. making cellphone call, putting an object down, and pointing to something, in complex scenes using a novel spatio-temporal descriptor based approach. A new spatio-temporal descriptor, which temporally integrates the statistics of a set of response maps of low-level features, e.g. image gradients and optical flows, in a space-time cube, is proposed to capture the characteristics of actions in terms of their appearance and motion patterns. Based on this kind of descriptors, the bag-of-words method is utilized to describe a human figure as a concise feature vector. Then, these features are employed to train SVM classifiers at multiple spatial pyramid levels to distinguish different actions. Finally, a Gaussian kernel based temporal filtering is conducted to segment the sequences of events from a video stream taking account of the temporal consistency of actions. The proposed approach is capable of tolerating spatial layout variations and local deformations of human actions due to diverse view angles and rough human figure alignment in complex scenes. Extensive experiments on the 50-hour video dataset of TRECVid 2008 event detection task demonstrate that our approach outperforms the well-known SIFT descriptor based methods and effectively detects video events in challenging real-world conditions.

Action recognition via restricted dense trajectories and spatio-temporal co-occurrence feature

Action Recognition Based on Object Tracking and Dense Trajectories

Action Recognition Based on Dense Trajectories and Human Detection

Human Action Recognition Using Deep Learning Methods.

Vision-Based Action Recognition Of Construction Workers Using Dense Trajectories

Human action recognition in videos based on dense trajectory selection

An Improved Interest Point Detector for Human Action Recognition

Human Action Recognition by Fast Dense Trajectories.

A Study of Relative Motion Point Trajectories for Action Recognition

Activity Recognition Using Dense Long-Duration Trajectories

Human Action Recognition Based on Deep Images and Dense Trajectories

Action Recognition Using Edge Trajectories and Motion Acceleration Descriptor

Human Action Recognition with Trajectory Based Covariance Descriptor in Unconstrained Videos

A Novel Trajectory-VLAD Based Action Recognition Algorithm for Video Analysis.

Learning Deep Trajectory Descriptor for Action Recognition in Videos Using Deep Neural Networks.

Motion keypoint trajectory and covariance descriptor for human action recognition

Robust Human Action Recognition Based on Spatio-Temporal Descriptors and Motion Temporal Templates

Detecting video events based on action recognition in complex scenes using spatio-temporal descriptor.

Action Recognition Using Trajectories of Joints

Trajectory-Based Modeling Of Human Actions With Motion Reference Points

Human Activity Recognition based on Dynamic Spatio-Temporal Relations