Hyunggic !

Vision-based Object Detection and Tracking

Detect and track multiple objects such as pedestrians, bicyclists, and vehicles using a monocular camera mounted on a moving vehicle ! Through a collaboration with General Motors, first half of my PhD work focused on a real-time implementation of vision-based object detection and tracking framework which can be applied to automotive safety applications. To this end, I have been developing a detection system based on Pedro Felzenszwalb's Deformable Part-based Models (with star-cascade algorithm) and a fast tracking system based on the Kalman filter. The system is well integrated into our new vehicle, Autonomous Cadillac SRX4, and shows a promising pedestrian/bicyclist detection performance and an amazing vehicle/motorcycle detection performance !

Our research group at CMU (a.k.a ADCRL, Autonomous Driving Collaboration Research Lab.) performed an amazing engineering work for building a beautiful self-driving vehicle as shown below. Some important aspects of the efforts were introduced in the following paper:

Towards a Viable Autonomous Driving Research Platform
Junqing Wei, Jarrod M. Snider, Junsung Kim, John M. Dolan, Raj Rajkumar and Bakhtiar Litkouhi
IEEE Intelligent Vehicle Symposium, 2013

[Camera and Computing Cluster Configuration of Our Autonomous SRX]

(a) Our Autonomous SRX. (b) Forward-looking Camera Setup (Monocular). (c) Computing Cluster and Control Console.

The following list of papers basically shows the history of its developement.

Real-Time Pedestrian and Vehicle Detection for Automotive Active Safety Systems
H. Cho, P. Rybski, S. W. Bang, W. Zhang, and B.V.K. Vijaya Kumar
IEEE Transaction on Intelligent Transportation Systems 2013 (under review), (project page).
Real-time Pedestrian Detection with Deformable Part Models
H. Cho, P. Rybski, A. Bar-Hillel, and W. Zhang
IEEE Intelligent Vehicles Symposium (IV 2012), Alcala de Henares, June, 2012, (project page).
Vision-based 3D Bicycle Tracking using Deformable Part Model and Interacting Multiple Model Filter
H. Cho, P. Rybski, and W. Zhang
IEEE International Conference on Robotics and Automation, 2011, Shanghai, China, 2011 (project page)
(Best Automation Paper Award - Finalist)
Vision-based Bicycle Detection and Tracking using a Deformable Part Model and an EKF Algorithm
H. Cho, P. Rybski, and W. Zhang
IEEE ITSC, 2010, Madeira Island, Portugal, 2010

[Selected Videos]

1. Pedestrian detection result on Caltech Pedestrian Dataset (IV 2012), set07_v000 with scene geometry

2. Vehicle detection result on Pittsburgh Dataset (ITS 2013, Showing the power of detection !)

3. Vehicle detection result on Pittsburgh Dataset (ITS 2013, Showing the power of tracking !)

4. Pedestrian detection with a rear-view camera (For GM project, with tracking)

Click here for a full list of publications