Unusual Event Detection in Crowded Scenes Using Bag of LBPs in Spatio-Temporal Patches

Abstract

Modelling events in densely crowded environments remains challenging, due to the diversity of events and the noise in the scene. We propose a novel approach for anomalous event detection in crowded scenes using dynamic textures described by the Local Binary Patterns from Three Orthogonal Planes (LBP-TOP) descriptor. The scene is divided into spatio-temporal patches where LBP-TOP based dynamic textures are extracted. We apply hierarchical Bayesian models to detect the patches containing unusual events. Our method is an unsupervised approach, and it does not rely on object tracking or background subtraction. We show that our approach outperforms existing state of the art algorithms for anomalous event detection in UCSD dataset.

Brief review on gender classification techniques

“…This system about has brought about 24% deduction of MAEs, as compared to the best results of most of the previous approaches. Xu et al (2011) claimed that it was very challenging to track all the individuals in a crowded scene. As a result, it is impractical to extract object trajectories as the feature to represent events in crowded scenes.…”

“…Xu et al (2011) gaussian mixture model (gmm) This is the fastest algorithm for learning mixture models.…”

Abnormality Detection with Deep Learning

“…These two detection approaches have been widely used in literature for visualising the outputs [5][6][7]35]. The abnormality in the video frames is visualised through frame-based detection whereas the locations of the abnormal patches in the frame can be detected through patch-based detection.…”

“…Jingxin et al, [5] in his paper proposed LBP-TOP (Linear Binary Pattern from Three Orthogonal Planes) for anomalous event detection in crowded scenes using dynamic textures. This approach is our baseline model where deep learning features will be applied over handcrafted features.…”

“…The feature extraction performed in [5] uses dynamic textures [34] described by the Local Binary Patterns from Three Orthogonal Planes (LBP-TOP) descriptor for anomalous event detection in crowded scenes. A local 3x3 kernel has been used for feature extraction.…”