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.

Interactive Surveillance Event Detection through Mid-level Discriminative Representation

Foreground Gating and Background Refining Network for Surveillance Object Detection

Intelligent Video Surveillance for Checking Attendance of Traffic Controllers in Level Crossing

A Comprehensive Solution for Detecting Events in Complex Surveillance Videos

A Unified Framework with a Benchmark Dataset for Surveillance Event Detection.

PKU-NEC @TRECVID2012 SED - Uneven-Sequence Based Event Detection in Surveillance Video.

A System Based On Sequence Learning For Event Detection In Surveillance Video

Early Event Detection Based on Dynamic Images of Surveillance Videos

A Framework for an Event Driven Video Surveillance System

Esur: A System for Events Detection in Surveillance Video

Robust Event Detection Scheme for Complex Scenes in Video Surveillance

TJUT-TJU@TRECVID 2011: Surveillance Event Detection.

Semi-supervised Early Event Detection.

A Representative-Based Framework For Parsing And Summarizing Events In Surveillance Videos

A Spatio-Temporal Deep Architecture for Surveillance Event Detection Based on ConvLSTM

Training-free Monocular 3D Event Detection System for Traffic Surveillance

Automatic Moving Object Detection and Indexing of Surveillance System

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

Surveillance Event Detection Based on CNN

Multi-object Events Recognition from Video Sequences Using Extended Finite State Machine

Bi-Level Semantic Representation Analysis for Multimedia Event Detection