Crowd Prediction, Tracking and Anomaly Detection

Learn about my crowd-related research at GAMMA


If you are having trouble streaming the YouTube video, here is the direct download link.
Realtime Pedestrian Behavior Learning for Path Prediction and Navigation
Realtime Pedestrian Behavior Learning for Path Prediction and Navigation

Realtime Pedestrian Behavior Learning for Path Prediction and Navigation

We present a real-time algorithm to learn varying pedestrian behavior characteristics from real-world videos. Our formulation is based on behavior classification using Personality Trait Theory. We present a scheme to dynamically learn the behavior of every pedestrian in the scene and use that to compute its motion parameters and future states using Bayesian inference. This behavior learning scheme makes no assumptions about prior pedestrian motion or crowd density and uses a precomputed database. We use behavior classification for real-time path prediction and navigation in low and medium density videos with tens of pedestrians. We highlight the improvement in accuracy (up to 24%) over prior prediction and navigation methods.

Project Link:
Realtime Pedestrian Behavior Learning for Path Prediction and Navigation - Submitted to IEEE International Conference on Robotics and Automation 2017 - Aniket Bera, Dinesh Manocha

Realtime Anomaly Detection Using Trajectory-Level Crowd Behavior Learning
Interactive and Adaptive Data-Driven Crowd Simulation

Realtime Anomaly Detection Using Trajectory-Level Crowd Behavior Learning

We present an algorithm for realtime anomaly detection in low to medium density crowd videos using trajectory-level behavior learning. Our formulation combines online tracking algorithms from computer vision, non-linear pedestrian motion models from crowd simulation, and Bayesian learning techniques to automatically compute the trajectory-level pedestrian behaviors for each agent in the video. These learned behaviors are used to segment the trajectories and motions of different pedestrians or agents and detect anomalies. We demonstrate the interactive performance on the PETS 2016 ARENA dataset as well as indoor and outdoor crowd video benchmarks consisting of tens of human agents.

Project Links:
Realtime Anomaly Detection Using Trajectory-Level Crowd Behavior Learning - Proceedings of CVPR 2016 Workshop - PETS - Aniket Bera, Dinesh Manocha

LCrowdV: Generating Labeled Videos for Simulation-based Crowd Behavior Learning
UNC Crowd Scene Analysis Dataset

LCrowdV: Generating Labeled Videos for Simulation-based Crowd Behavior Learning

We present a novel procedural framework to generate an arbitrary number of labeled crowd videos (LCrowdV). The resulting crowd video datasets are used to design accurate algorithms or training models for crowded scene understanding. Our overall approach is composed of two components: a procedural simulation framework for generating crowd movements and behaviors, and a procedural rendering framework to generate different videos or images. Each video or image is automatically labeled based on the environment, number of pedestrians, density, behavior (agent personality), flow, lighting conditions, viewpoint, noise, etc. Furthermore, we can increase the realism by combining synthetically-generated behaviors with real-world background videos. We demonstrate the benefits of LCrowdV over prior lableled crowd datasets, by augmenting real dataset with it and improving the accuracy in pedestrian detection. LCrowdV has been made available as an online resource.

Project Link:
LCrowdV: Generating Labeled Videos for Simulation-based Crowd Behavior Learning Ernest Cheung, Wong Tsan Kwong, Aniket Bera, Dinesh Manocha

Interactive and Adaptive Data-Driven Crowd Simulation
Interactive and Adaptive Data-Driven Crowd Simulation

Interactive and Adaptive Data-Driven Crowd Simulation

We present an adaptive data-driven algorithm for interactive crowd simulation. Our approach combines realistic trajectory behaviours extracted from videos with synthetic multi-agent algorithms to generate plausible simulations. We use statistical techniques to compute the movement patterns and motion dynamics from noisy 2D trajectories extracted from crowd videos. We also present results from preliminary user studies that evaluate the trajectory behaviours generated by our algorithm.

Project Links:
Interactive and Adaptive Data-Driven Crowd Simulation - Proceedings of VR IEEE 2016 (To Appear) - Sujeong Kim, Aniket Bera, Dinesh Manocha

Crowd Content Generation using Behavior Learning
Interactive Crowd Content Generation and Analysis using Trajectory-level Behavior Learning

Interactive Crowd Content Generation and Analysis using Trajectory-level Behavior Learning

We present a trajectory extraction and behavior-learning algorithm for data-driven crowd simulation. Our formulation is based on incrementally learning pedestrian motion models and behaviors from crowd videos. We combine this learned crowd-simulation model with an online tracker based on particle filtering to compute accurate, smooth pedestrian trajectories. We refine this motion model using an optimization technique to estimate the agents' simulation parameters.

Project Link:
Interactive Crowd Content Generation and Analysis using Trajectory-level Behavior Learning - Proceedings of IEEE International Symposium on Multimedia 2015 (To Appear) - Sujeong Kim, Aniket Bera, Dinesh Manocha

Realtime Pedestrian Path Prediction using Global and Local Movement Patterns
Realtime Pedestrian Path Prediction using Global and Local Movement Patterns

GLMP - Realtime Pedestrian Path Prediction using Global and Local Movement Patterns

We present a novel real-time algorithm to predict the path of pedestrians in cluttered environments. Our approach makes no assumption about pedestrian motion or crowd density, and is useful for short-term as well as long-term prediction. We interactively learn the characteristics of pedestrian motion and movement patterns from 2D trajectories using Bayesian inference. These include local movement patterns corresponding to the current and preferred velocities and global characteristics such as entry points and movement features.

Project Link:
GLMP- Realtime Pedestrian Path Prediction using Global and Local Movement Patterns - Proceedings of IEEE International Conference on Robotics and Automation 2016 - Aniket Bera, Dinesh Manocha

Real-time Adaptive Pedestrian Tracking
Real-time Adaptive Pedestrian Tracking

Real-time Adaptive Pedestrian Tracking

We present a novel, realtime algorithm to compute the trajectory of each pedestrian in moderately dense crowd scenes. Our formulation is based on an adaptive particle filtering scheme that uses a multi-agent motion model based on velocity-obstacles, and takes into account local interactions as well as physical and personal constraints of each pedestrian.

Project Links:
Realtime Multilevel Crowd Tracking using Reciprocal Velocity Obstacles - Proceedings of IEEE International Conference on Pattern Recognition 2014 - Aniket Bera, Dinesh Manocha
AdaPT: Real-time Adaptive Pedestrian Tracking for crowded scenes - Proceedings of IEEE International Conference on Robotics and Automation 2014 - Aniket Bera, Nico Galoppo, Dillon Sharlet, Adam Lake, Dinesh Manocha

Parameter Learning for Data-Driven Crowd Simulation
Parameter Learning for Data-Driven Crowd Simulation

Parameter Learning for Data-Driven Crowd Simulation

We present a trajectory extraction and behavior-learning algorithm for data-driven crowd simulation. Our formulation is based on incrementally learning pedestrian motion models and behaviors from crowd videos. We combine this learned crowd-simulation model with an online tracker based on particle filtering to compute accurate, smooth pedestrian trajectories. We refine this motion model using an optimization technique to estimate the agents' simulation parameters.

Project Link:
Online parameter learning for data-driven crowd simulation and content generation - Computers & Graphics (Journal) - Aniket Bera, Sujeong Kim, Dinesh Manocha
Efficient Trajectory Extraction and Parameter Learning for Data-Driven Crowd Simulation - Proceedings of Graphics Interface 2015 - Aniket Bera, Sujeong Kim, Dinesh Manocha

Crowd Tracking using Hybrid Discrete + Continuum Model
Crowd Tracking using Hybrid Discrete + Continuum Model

Crowd Tracking using a Hybrid Discrete + Continuum Model

In order to improve the tracking accuracy, we use a hybrid motion model that combines discrete and continuous flow models. The discrete model is based on microscopic agent formulation and is used for local navigation, interaction, and collision avoidance. The continuum model accounts for macroscopic behaviors, including crowd orientation and flow.

Project Link:
REACH - Realtime Crowd tracking using a Hybrid motion model - Proceedings of IEEE International Conference on Robotics and Automation 2015 (To Appear) - Aniket Bera, Dinesh Manocha

Acknowledgements

Our collaborators and funding agencies.

To get in touch with the authors, please contact the authors (Aniket Bera and Dinesh Manocha) at ab@cs.unc.edu

Crowd Research

To see more work on motion and crowd simulation models in our GAMMA group, visit - http://gamma.cs.unc.edu/research/crowds/