Part-based multiple-person tracking with partial occlusion handling (original) (raw)
Related papers
Online multiple people tracking-by-detection in crowded scenes
7'th International Symposium on Telecommunications (IST'2014), 2014
Multiple people detection and tracking is a challenging task in real-world crowded scenes. In this paper, we have presented an online multiple people tracking-by-detection approach with a single camera. We have detected objects with deformable part models and a visual background extractor. In the tracking phase we have used a combination of support vector machine (SVM) person-specific classifiers, similarity scores, the Hungarian algorithm and inter-object occlusion handling. Detections have been used for training person-specific classifiers and to help guide the trackers by computing a similarity score based on them and spatial information and assigning them to the trackers with the Hungarian algorithm. To handle inter-object occlusion we have used explicit occlusion reasoning. The proposed method does not require prior training and does not impose any constraints on environmental conditions. Our evaluation showed that the proposed method outperformed the state of the art approaches by 10% and 15% or achieved comparable performance
Person tracking with partial occlusion handling
2015 IEEE Workshop on Signal Processing Systems (SiPS), 2015
Occlusion is a challenge for tracking especially in dynamic scene. It adds the consideration for background modeling. In the condition, the tracker will be influenced by both occlusions and background. In this paper, we address the problem by proposing a robust algorithm based on improved particle filter using discriminative model without background modeling. Discriminative model offers accurate templates for occlusion detection by alleviating influence from background pixels. Since particle filter cannot carry out effective tracking under heavy occlusion, blocking is introduced to solve the problem by abandoning unobservable parts of the target. Experimental results show that our algorithm can work persistently and effectively under severe occlusion even in dynamic scene compared with state-of-the-arts.
Improved Tracking of Multiple Humans with Trajectory Prediction and Occlusion Modeling
1998
A combined 2D, 3D approach is presented that allows for robust tracking of moving bodies in a given environment as observed via a single, uncalibrated video camera. Lowlevel features are often insufficient for detection, segmentation, and tracking of non-rigid moving objects. Therefore, an improved mechanism is proposed that combines lowlevel (image processing) and mid-level (recursive trajectory estimation) information obtained during the tracking process. The resulting system can segment and maintain the tracking of moving objects before, during, and after occlusion. At each frame, the system also extracts a stabilized coordinate frame of the moving objects. This stabilized frame can be used as input to motion recognition modules. The approach enables robust tracking without constraining the system to know the shape of the objects being tracked beforehand; although, some assumptions are made about the characteristics of the shape of the objects, and how they evolve with time. Experiments in tracking moving people are described.
Multimedia Tools and Applications, 2016
Although much progress has been made in multi-object tracking in recent decades due to its variety of applications including visual surveillance, traffic monitoring and medical image analysis, some difficult challenges such as the variation of object appearance and partial occlusion are still going on. In this work, we propose an effective multi-human tracking system called part-based appearance modelling and grouping-based tracklet association-based multi-human tracking (PAM-GTA-MHT). The proposed appearance model based on the upper body-centered multi-view human body part model can effectively resolve the drawback caused by inter-object occlusions and low camera positions. The grouping method embedded in global tracklet association can improve discriminability among targets with similar appearances when they are located sufficiently far away from each other. Thus, there is no need to compare all possible pairs of the detected targets in the tracklet association stage and thus it has the potential to enhance the tracking speed. We quantitatively evaluated the performance of our proposed approach on four challenging publicly available datasets and achieved a significant improvement compared to the state-of-the-art methods.
Occluded Human Tracking and Identification Using Image Annotation
2012
Abstract The important task of human tracking can be difficult to implement in real world environment as the videos can involve complex scenes, severe occlusion and even moving background. Tracking individual objects in a cluttered scene is an important aspect of surveillance. In addition, the systems should also avoid misclassification which can lead to inaccurate tracking. This paper makes use of an efficient image annotation for human tracking.
Online Multiperson Tracking-by-Detection from a Single, Uncalibrated Camera
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000
In this paper, we address the problem of automatically detecting and tracking a variable number of persons in complex scenes using a monocular, potentially moving, uncalibrated camera. We propose a novel approach for multi-person tracking-bydetection in a particle filtering framework. In addition to final high-confidence detections, our algorithm uses the continuous confidence of pedestrian detectors and online trained, instance-specific classifiers as a graded observation model. Thus, generic object category knowledge is complemented by instance-specific information. The main contribution of this paper is to explore how these unreliable information sources can be used for robust multi-person tracking. The algorithm detects and tracks a large number of dynamically moving persons in complex scenes with occlusions, does not rely on background modeling, requires no camera or ground plane calibration, and only makes use of information from the past. Hence, it imposes very few restrictions and is suitable for online applications. Our experiments show that the method yields good tracking performance in a large variety of highly dynamic scenarios, such as typical surveillance videos, webcam footage, or sports sequences. We demonstrate that our algorithm outperforms other methods that rely on additional information. Furthermore, we analyze the influence of different algorithm components on the robustness.
Tracking people under heavy occlusions by layered data association
Multimedia Tools and Applications, 2014
One of the main difficulties in video tracking of people arises in scenarios where targets are repeatedly and extensively occluded by other moving objects. These types of occlusions significantly affect the measurements of the person's position, motion, shape and appearance, posing major challenges to correct tracking and data association. In this paper, we present a method for tracking people in videos based on a simplified part-based model only loosely associated with body parts. Data association is provided by a layered data association approach which performs association at feature, part and global levels in a hierarchical fashion. Occlusions are detected and managed at the part level, with corresponding model update strategies. In addition, the tracker does not make any assumption on the target's motion direction, thus allowing tracking to withstand abrupt sideways movements and changes of directions that frequently occur in busy scenes. Experimental results against popular trackers such as mean shift, particle filters and the recent k-shortest paths (KSP) tracker based on a variety of performance indicators and datasets including ETISEO, AVSS 2007 and PETS 2009 show the effectiveness of the proposed tracker.
Multiple People Detection and Tracking
Mantech Publications , 2021
Multiple people detection in real-time is still a challenging task despite having different techniques. It is challenging because partially occluded people are still often not recognized in a heavily populated area, and also due to Non-Maximum suppression, correct bounding boxes are also discarded, which leads to imprecision in the detections. This paper presents the various modifications done to multiple people detection and tracking algorithms, which improves the efficiency and accuracy of the previously used cases.