International Journal of Computer Vision
Matching Shape Sequences in Video with Applications in Human Movement Analysis
IEEE Transactions on Pattern Analysis and Machine Intelligence
Activity Awareness: from Predefined Events to New Pattern Discovery
ICVS '06 Proceedings of the Fourth IEEE International Conference on Computer Vision Systems
Modeling, Clustering, and Segmenting Video with Mixtures of Dynamic Textures
IEEE Transactions on Pattern Analysis and Machine Intelligence
Video sequence querying using clustering of objects' appearance models
ISVC'07 Proceedings of the 3rd international conference on Advances in visual computing - Volume Part II
Activity recognition via classification constrained diffusion maps
ISVC'06 Proceedings of the Second international conference on Advances in Visual Computing - Volume Part I
Hi-index | 0.00 |
Video surveillance of large facilities, such as airports, rail stations and casinos, is developing rapidly. Cameras installed at such locations often overlook large crowds, which makes problems such as activity and scene understanding very challenging. Traditional activity recognition methods which rely on input from lower level processing units dealing with background subtraction and tracking, are unable to cope with frequent occlusions in such scenes. In this paper, we propose a novel activity representation and recognition method that bypasses these commonly used low level modules. We model each local spatio-temporal patch as a dynamic texture. Using a Martin distance metric to compare two patches based on their estimated dynamic texture parameters, we present a method to temporally stitch together local regions to form activity streamlines and represent each streamline by its constituent dynamic textures. This allows us to seamlessly perform activity recognition without explicitly detecting individuals in the scene. We demonstrate our method on multiple real data sets and show promising results.