• Corpus ID: 18732116

Fusion of Multiple Cues from Color and Depth Domains using Occlusion Aware Bayesian Tracker

  title={Fusion of Multiple Cues from Color and Depth Domains using Occlusion Aware Bayesian Tracker},
  author={Meshgi Kourosh and Maeda Shin-ichi and Oba Shigeyuki and Ishii Shin},
  journal={IEICE technical report. Neurocomputing},
Occlusion Aware Bayesian Tracker Kourosh MESHGI Shin-ichi MAEDA Shigeyuki OBA and Shin ISHII †Graduate School of Informatics, Kyoto University, Gokasho, Uji-shi, Kyoto, 611-0011 Japan E-mail: †{meshgi-k,ichi,oba,ishii}@sys.i.kyoto-u.ac.jp Abstract Object tracking has attracted considerable attention recently because of high demands for its everyday-life applications. Appearance-based trackers had a significant improvement during the last decade, however they are still struggling with some… 

Figures and Tables from this paper

Siamese Network Object Tracking Algorithm Based on Squeeze-and-Excitation
A Siamese Network tracker based on SE-ResNet-50 is proposed and is much better than some popular trackers in solving the problems of occlusion, illumination change and object deformation.
Multi-view task-driven recognition in visual sensor networks
This paper converts the problem of visual analysis in resource-limited scenarios to a multiview representation learning, and shows that the key to finding properly compressed representation is to exploit the position of cameras with respect to each other as a norm-based regularization in the particular signal representation of sparse coding.
Descriptor-based tracking algorithm using a depth camera
A new descriptor-based algorithm for real-time object tracking using the information from a Microsoft Kinect depth camera is proposed, using histograms of oriented gradients calculated from the circular sliding regions of the scene image as a descriptor for the object tracked.


Enhancing Probabilistic Appearance-Based Object Tracking with Depth Information: Object Tracking under Occlusion
This study proposed a consistent way of integrating color and depth information in a particle filter framework to efficiently perform the tracking task, but also enhanced the previous color-based particle filtering to achieve trajectory independence and consistency with respect to the target scale.
BraMBLe: a Bayesian multiple-blob tracker
  • M. Isard, J. MacCormick
  • Computer Science
    Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001
  • 2001
A multi-blob likelihood function which assigns directly comparable likelihoods to hypotheses containing different numbers of objects and a Bayesian filter for tracking multiple objects when the number of objects present is unknown and varies over time are introduced.
Monocular 3D scene understanding with explicit occlusion reasoning
This paper proposes an approach that tightly integrates monocular 3D scene tracking-by-detection with explicit object-object occlusion reasoning and demonstrates that this approach yields more robust detection and tracking of partially visible pedestrians, even when they are occluded over long periods of time.
Tracking Revisited Using RGBD Camera: Unified Benchmark and Baselines
A unified benchmark dataset of 100 RGBD videos with high diversity is constructed, different kinds of RGBD tracking algorithms using 2D or 3D model are proposed, and a quantitative comparison of various algorithms with RGB or RGBD input is presented.
Learning to Detect and Track People in RGBD Data
This paper addresses the novel problem of detecting people in RGB-D data in distances far beyond the recommended sensor operating range (called adequate play space), and proposes an online-learning method of target appearances along with its integration into the MHT.
An adaptive color-based particle filter
Detecting and tracking people using an RGB-D camera via multiple detector fusion
A system for detecting and tracking people from image and depth sensors on board a mobile robot that combines an ensemble of detectors in a unified framework, is efficient, and has the potential to incorporate multiple sensor inputs.
People detection in RGB-D data
This paper takes inspiration from the Histogram of Oriented Gradients (HOG) detector to design a robust method to detect people in dense depth data, called HOD, and proposes Combo-HOD, a RGB-D detector that probabilistically combines HOD and HOG.
Expanding histogram of colors with gridding to improve tracking accuracy
  • Kourosh Meshgi, S. Ishii
  • Computer Science
    2015 14th IAPR International Conference on Machine Vision Applications (MVA)
  • 2015
The idea of gridding for color histogram is proposed, which grants specific statistical property to the histogram through a decomposition phase followed by a recombination stage, and a thorough comparison of the modern similarity functions and model update techniques in RGB colorspace reveals that this method in combination with established similarity measures, enhances the tracking performance.
Real-time tracking of non-rigid objects using mean shift
The theoretical analysis of the approach shows that it relates to the Bayesian framework while providing a practical, fast and efficient solution for real time tracking of non-rigid objects seen from a moving camera.