Real-time people detection and tracking for indoor surveillance using multiple top-view depth cameras

Tseng, Ting-En; Liu, An-Sheng; Hsiao, Po-Hao; Huang, Cheng-Ming; Fu, Li‐Chen

doi:10.1109/iros.2014.6943136

Cited by 37 publications

(25 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Early research with top-view cameras focused on techniques for tracking people's position [20,42,47,50]. To address the lack of support for top-view human pose estimation for interactive tabletops, Haubner et al [16] suggested to adopt Kinect's skeleton training pipeline [45] with a new dataset, where the subjects would wear color suits that have a distinct color-coding for each body part.…”

Section: Human Detection and Trackingmentioning

confidence: 99%

“…This setup requires a second depth-sensing camera to project the skeleton joint positions from the side-view coordinate system to the top-view coordinate system. Lin et al [30] extended the top-view person detector from [50] to recognize six different activities.…”

Section: Human Activity Recognitionmentioning

confidence: 99%

“…This simple mechanism allows the tracking system to run continuously without recalibration. To reduce depth-sensing noise from the camera and background subtraction, we smooth the foreground depth map using a 5 by 5 median filter, a common technique employed by previous work [44,50]. We keep the infrared map unfiltered to retain as much sparse infrared pixels on devices (which the system attempts to recognize) as possible, especially on small devices like mobile phones.…”

Section: Phase A: Top-view Depth-sensing Trackingmentioning

confidence: 99%

See 2 more Smart Citations

EagleSense

Houben

Marquardt

2017

Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

Real-time tracking of people's location, orientation and activities is increasingly important for designing novel ubiquitous computing applications. Top-view camera-based tracking avoids occlusion when tracking people while collaborating, but often requires complex tracking systems and advanced computer vision algorithms. To facilitate the prototyping of ubiquitous computing applications for interactive spaces, we developed EagleSense, a real-time human posture and activity recognition system with a single top-view depthsensing camera. We contribute our novel algorithm and processing pipeline, including details for calculating silhouetteextremities features and applying gradient tree boosting classifiers for activity recognition optimized for top-view depth sensing. EagleSense provides easy access to the real-time tracking data and includes tools for facilitating the integration into custom applications. We report the results of a technical evaluation with 12 participants and demonstrate the capabilities of EagleSense with application case studies.

show abstract

Section: Human Detection and Trackingmentioning

confidence: 99%

Section: Human Activity Recognitionmentioning

confidence: 99%

Section: Phase A: Top-view Depth-sensing Trackingmentioning

confidence: 99%

See 1 more Smart Citation

EagleSense

Houben

Marquardt

2017

Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

show abstract

“…[28] reviews such methods), but only a few works address the problem with depth camera networks. The method presented in [27] employs the plan-view approach for depth-based people tracking. In the approach, the cameras are aligned to a common top-view coordinate system using pairwise affine transformations, and one large height map of the scene is constructed.…”

Section: Introductionmentioning

confidence: 99%

Auto-calibration of depth camera networks for people tracking

Korkalo

Tikkanen

Kemppi

et al. 2019

Machine Vision and Applications

View full text Add to dashboard Cite

We address the problem of calibrating an embedded depth camera network designed for people tracking purposes. In our system, the nodes of the network are responsible for detecting the people moving in their view, and sending the observations to a centralized server for data fusion and tracking. We employ a plan-view approach where the depth camera views are transformed to top-view height maps where people are observed. As the server transforms the observations to a global planview coordinate system, accurate geometric calibration of the sensors has to be performed. Our main contribution is an auto-calibration method for such depth camera networks. In our approach, the sensor network topology and the initial 2D rigid transformations that map the observations to the global frame are determined using observations only. To distribute the errors in the initial calibration, the transformation parameters and the estimated positions of people are refined using a global optimization routine. To overcome inaccurate depth camera parameters, we re-calibrate the sensors using more flexible transformations, and experiment with similarity, affine, homography and thin-plate spline mappings. We evaluate the robustness, accuracy and precision of the approach using several real-life data sets, and compare the results to a checkerboardbased calibration method as well as to the ground truth trajectories collected with a mobile robot.

show abstract

“…Although Kinect would provide reliable measurements, its measurable range is shorter than that of cameras, and it is hard to apply in outdoor environments. Tseng also applies Kinect; however, the sensor is hung from the ceiling [4]. This approach can avoid the occlusion problem; therefore, it is computationally efficient.…”

Section: Introductionmentioning

confidence: 99%

Complementary human detection and multiple feature based tracking using a stereo camera

2017

View full text Add to dashboard Cite

A human detection and tracking method using a stereo camera is presented in this paper. Two human detection methods are independently implemented, and the results are combined to reduce misdetection and nondetection. The tracking step is based on particle filtering. In the data association phase, we introduce three features: distance, traveling direction, and color. The color feature is obtained from every segmented detection window. Eligibility and co-occurrence of the blocks provide robustness to occlusion. The proposed method is tested by using measurement data at the entrance of a building, where occlusion is frequently observed. The experiments demonstrate improved tracking performance over standard particle filtering.

show abstract

Real-time people detection and tracking for indoor surveillance using multiple top-view depth cameras

Cited by 37 publications

References 17 publications

EagleSense

EagleSense

Auto-calibration of depth camera networks for people tracking

Complementary human detection and multiple feature based tracking using a stereo camera

Contact Info

Product

Resources

About