CONDENSATION—Conditional Density Propagation forVisual Tracking
International Journal of Computer Vision
Color-Based Probabilistic Tracking
ECCV '02 Proceedings of the 7th European Conference on Computer Vision-Part I
An Experimental Comparison of Min-Cut/Max-Flow Algorithms for Energy Minimization in Vision
IEEE Transactions on Pattern Analysis and Machine Intelligence
Analysis of Persistent Motion Patterns Using the 3D Structure Tensor
WACV-MOTION '05 Proceedings of the IEEE Workshop on Motion and Video Computing (WACV/MOTION'05) - Volume 2 - Volume 02
Pedestrian Behavior at Bottlenecks
Transportation Science
Floor Fields for Tracking in High Density Crowd Scenes
ECCV '08 Proceedings of the 10th European Conference on Computer Vision: Part II
PLEdestrians: a least-effort approach to crowd simulation
Proceedings of the 2010 ACM SIGGRAPH/Eurographics Symposium on Computer Animation
Large Displacement Optical Flow: Descriptor Matching in Variational Motion Estimation
IEEE Transactions on Pattern Analysis and Machine Intelligence
Visual crowd surveillance through a hydrodynamics lens
Communications of the ACM
A comparative study of energy minimization methods for markov random fields
ECCV'06 Proceedings of the 9th European conference on Computer Vision - Volume Part II
Tracking Pedestrians Using Local Spatio-Temporal Motion Patterns in Extremely Crowded Scenes
IEEE Transactions on Pattern Analysis and Machine Intelligence
Hidden Markov Model on a unit hypersphere space for gesture trajectory recognition
Pattern Recognition Letters
Tracking in dense crowds using prominence and neighborhood motion concurrence
Image and Vision Computing
| Hi-index | 0.00 |
Video analysis of crowded scenes is challenging due to the complex motion of individual people in the scene. The collective motion of pedestrians form a crowd flow, but individuals often largely deviate from it as they anticipate and react to each other. Deviations from the crowd decreases the pedestrian's efficiency: a sociological concept that measures the difference of actual motion from the intended speed and direction. In this paper, we derive a novel method for estimating pedestrian efficiency from videos. We first introduce a novel crowd motion model that encodes the temporal evolution of local motion patterns represented with directional statistics distributions. This model is then used to estimate the intended motion of pedestrians at every space-time location, which enables visual measurement of the pedestrian efficiency. We demonstrate the use of this pedestrian efficiency to detect unusual events and to track individuals in crowded scenes. Experimental results show that the use of pedestrian efficiency leads to state-of-the-art accuracy in these critical applications.