Search 250 million research papers

Multiple Object Tracking

Aiyu Cui · 2018

Cited by 0

Multiple-Object Tracking

Kathryn Barker, Roy Allen, Peter McGeorge · Experimental Psychology (formerly Zeitschrift für Experimentelle Psychologie) · 2009

Research has demonstrated that individuals who routinely engage in complex visuospatial tasks (e.g., radar operators) show an enhanced ability to track multiple randomly moving targets. This study examined tracking expertise using members of a University Officer Training Corps (OTCs) who regularly engage in tasks requiring good dynamic spatial cognition. As expected, the results show that OTCs have enhanced tracking ability relative to other undergraduates. More importantly, they support the idea that, while one set of executive processes are involved in the moment-by-moment updating of the visuospatial representations necessary for dynamic, multiple-object tracking, other processes are activated when whole object sets disappear simultaneously, to create a long-term memory trace of the objects' locations at the moment of their disappearance. Expertise only arose in the former processes, but was lost after a short decay period, such as occurred with a delayed response.

Cited by 23

Multiple-Object Tracking

K. Botterill, Rachel Grasmeder Allen, Peter McGeorge · Experimental Psychology (formerly Zeitschrift für Experimentelle Psychologie) · 2010

The Multiple-Object Tracking paradigm has most commonly been utilized to investigate how subsets of targets can be tracked from among a set of identical objects. Recently, this research has been extended to examine the function of featural information when tracking is of objects that can be individuated. We report on a study whose findings suggest that, while participants can only hold featural information for roughly two targets this task does not affect tracking performance detrimentally and points to a discontinuity between the cognitive processes that subserve spatial location and featural information.

Cited by 18

Multiple object tracking

Alex O. Holcombe · WikiJournal of Science · 2023

In psychology and neuroscience, multiple object tracking (MOT) refers to the ability of humans and other animals to simultaneously monitor multiple objects as they move. It is also the term for a laboratory technique used to study this ability. In an MOT study, a number of identical moving objects are presented on a display. Some of the objects are designated as targets while the rest serve as distractors. Study participants try to monitor the changing positions of the targets as they and the distractors move about. At the end of the trial, participants typically are asked to indicate the final positions of the targets. The results of MOT experiments have revealed dramatic limitations on humans' ability to simultaneously monitor multiple moving objects. For example, awareness of features such as color and shape is disrupted by the objects' movement.

Cited by 1Open access

Multiple Object Tracking

Takács, Ákos · Brno University of Technology Digital Library (Brno University of Technology) · 2009

Práce se zabývá teoretickým popisem, návrhem a implementací systému pro sledování objektů v zorném poli kamery pomocí jednoduché webkamery. Při zpracování obrazových dat je použitá knihovna OpenCV. Systém je schopný řídit otáčení kamery tak, aby uživatelem vybraný objekt byl v středu obrazu. Pro řízení kamery je využívána řídící deska FITkit. Program je implementován v jazyce C++.

Cited by 0

Multiple object tracking

Qiaokang Liang, Hai Qin, Shao Xiang · Elsevier eBooks · 2025

Cited by 0

Multiple object tracking

Yan Feng · Tampere University Institutional Repository (Tampere University) · 2020

Multiple object tracking, a middle-level task, is a critical foundation to support advanced research activities, like pose analysis or motion recognition. In this thesis, the relationship between object detection, single-object tracking, and multiple object tracking was explored and discussed.
\nOn this basis, the Single Shot MultiBox Detector (SSD), SiamMask network, and Detect and Track (D&T) model have been utilized, modified and evaluated. 
\nD&T model is the offline Detection Based Tracking (DBT) network. We had observed the benefits of this correlation loss application via researching on D&T model, and we trained the D&T network on the dataset combination containing the person objects so as to make it useful in reality. In addition, SSD had been applied as the detector with the same tracking methods, during which process the effect of the diverse detectors on the MOT experiments could be figured out. The last experiment was executed on the Single Object Tracking (SOT) model-SiamMask. The original SiamMask network is an online Detection Free Tracking (DFT) network. In order to adapt to the situation of multi-target tracking, it had been modified to initialize multiple target objects with an SSD detector at every specific interval.
\nHaving prepared all the multiple object tracking models, we carried out the evaluation with the MOT17DET dataset. The evaluation metrics for multiple object tracking provided us with a standard view of their performance. In this procedure, we also obtained some helpful knowledge and experience for future MOT improvement.

Cited by 2Open access

Object-Centric Multiple Object Tracking

Zixu Zhao, Jiaze Wang, Max Horn et al. · arXiv (Cornell University) · 2023

Unsupervised object-centric learning methods allow the partitioning of scenes into entities without additional localization information and are excellent candidates for reducing the annotation burden of multiple-object tracking (MOT) pipelines. Unfortunately, they lack two key properties: objects are often split into parts and are not consistently tracked over time. In fact, state-of-the-art models achieve pixel-level accuracy and temporal consistency by relying on supervised object detection with additional ID labels for the association through time. This paper proposes a video object-centric model for MOT. It consists of an index-merge module that adapts the object-centric slots into detection outputs and an object memory module that builds complete object prototypes to handle occlusions. Benefited from object-centric learning, we only require sparse detection labels (0%-6.25%) for object localization and feature binding. Relying on our self-supervised Expectation-Maximization-inspired loss for object association, our approach requires no ID labels. Our experiments significantly narrow the gap between the existing object-centric model and the fully supervised state-of-the-art and outperform several unsupervised trackers.

Cited by 1Open access

Improving Multiple Object Tracking with Single Object Tracking

Linyu Zheng, Ming Tang, Yingying Chen et al. · 2021

Despite considerable similarities between multiple object tracking (MOT) and single object tracking (SOT) tasks, modern MOT methods have not benefited from the development of SOT ones to achieve satisfactory performance. The major reason for this situation is that it is inappropriate and inefficient to apply multiple SOT models directly to the MOT task, although advanced SOT methods are of the strong discriminative power and can run at fast speeds.In this paper, we propose a novel and end-to-end trainable MOT architecture that extends CenterNet by adding an SOT branch for tracking objects in parallel with the existing branch for object detection, allowing the MOT task to benefit from the strong discriminative power of SOT methods in an effective and efficient way. Unlike most existing SOT methods which learn to distinguish the target object from its local backgrounds, the added SOT branch trains a separate SOT model per target online to distinguish the target from its surrounding targets, assigning SOT models the novel discrimination. Moreover, similar to the detection branch, the SOT branch treats objects as points, making its online learning efficient even if multiple targets are processed simultaneously. Without tricks, the proposed tracker achieves MOTAs of 0.710 and 0.686, IDF1s of 0.719 and 0.714, on MOT17 and MOT20 benchmarks, respectively, while running at 16 FPS on MOT17.

Cited by 116

Multiple Object Tracking Paradigm

Maolin Ye · 2023

Cited by 0

Multiple Planar Object Tracking

Zhicheng Zhang, Shengzhe Liu, Jufeng Yang · 2023

Tracking both location and pose of multiple planar objects (MPOT) is of great significance to numerous real-world applications. The greater degree-of-freedom of planar objects compared with common objects makes MPOT far more challenging than well-studied object tracking, especially when occlusion occurs. To address this challenging task, we are inspired by amodal perception that humans jointly track visible and invisible parts of the target, and propose a tracking framework that unifies appearance perception and occlusion reasoning. Specifically, we present a dual-branch network to track the visible part of planar objects, including vertexes and mask. Then, we develop an occlusion area localization strategy to infer the invisible part, i.e., the occluded region, followed by a two-stream attention network finally refining the prediction. To alleviate the lack of data in this field, we build the first large-scale benchmark dataset, namely MPOT-3K. It consists of 3,717 planar objects from 356 videos and contains 148,896 frames together with 687,417 annotations. The collected planar objects have 9 motion patterns and the videos are shot in 6 types of indoor and outdoor scenes. Extensive experiments demonstrate the superiority of our proposed method on the newly developed MPOT-3K as well as other two popular single planar object tracking datasets. The code and MPOT-3K dataset are released on https://zzcheng.top/MPOT.

Cited by 6

Multiple Object Tracking Paradigm

Maolin Ye · 2024

Cited by 0

Multiple object tracking task.

Kameron K. Clayton (3158676), Jayaganesh Swaminathan (3158679), Arash Yazdanbakhsh (764064) et al. · Figshare · 2016

<p>(A) Cue phase: Target dots, between 1 to 5, are marked in green for 2 seconds to designate them as targets for the tracking task. (B) Tracking phase: target dots turn back to grey. All of the dots, now identical, move around randomly on the display for 7 seconds. (C) Response phase: subjects report the final locations of the target dots and receive feedback on correct and incorrect choices (grey dot with red outline = incorrect selection, green dot with black outline = target dot but not selected, and green dot with red outline = target dot and selected).</p>

Cited by 0Open access

praveen-palanisamy/multiple-object-tracking-lidar: Multiple-Object-Tracking-from-Point-Clouds_v1.0.2

Palanisamy, Praveen · Zenodo (CERN European Organization for Nuclear Research) · 2019

Multiple objects detection, tracking and classification from LIDAR scans/point-clouds v1.0.2 (2019-12-01) Added link to wiki pages Updated readme with suuported pointcloud sources Updated README to clarify real, sim, dataset LiDAR data Contributors: Praveen Palanisamy See README for more info.

Cited by 1Open access

iMOT: interactive Multiple Object Tracking

Ian M. Thornton, Todd S. Horowitz, HH Bülthoff · Journal of Vision · 2011

We introduce a novel, iPad-based experimental paradigm to study interactive multiple object tracking (MOT). In standard MOT, participants passively track a set of independently moving objects. We devised a new task in which participants interact with objects, rather than simply track them. We combined a typical MOT display with the path-guidance, touch-screen interface popularised in smart phone games such as Flight Controller and Harbor Master. We ran the experiment using a custom iPad application. A variable number of identical spheres (1.16° visual angle) moved slowly (0.89°/s) on independent linear trajectories within the full iPad display area (22.6° × 17°). Random direction changes occurred at intervals between 4 and 7 s. The participant's goal was to avoid object collisions by manually altering the sphere trajectories. This was achieved by touching the sphere and drawing a short linear or curved path away from it. This action created a visible path, which the sphere followed before resuming default random motion. In this initial study, we constrained path length, in order to prevent participants from creating complex “holding patterns”. A trial ended when the participant successfully avoided collisions for 30 s or when a collision occurred. A sphere was added to the display after a successful trial and removed after a collision, subject to a 6-sphere minimum. Twenty-four participants (ages 18 to 33, mean 23.58, sd 3.45); twelve from Korea University, twelve from Swansea University) completed 30 trials. There were no significant effects of culture, sex or previous gaming experience. The mean number of spheres participants could control was 8.39 (s.e.m. 0.14). This is consistent with estimates of passive tracking capacity at slow speeds. We hypothesize that executive control of scant attentional resources is critical for the current task. In support of this hypothesis, interactive tracking capacity actually increased with age (p = .016).

Cited by 2Open access

Attentional Based Multiple-Object Tracking

Mark Calafut · 2013

This paper investigates the attentional based tracking framework of Bazzani et al. (2011) and the general performance of attentional based tracking systems using different classification, prediction, and gaze selection techniques. Three object classifiers were implemented using the KNN technique, support vector machines, and ADA Boosting to recognize digits from the MNIST dataset. The performance of the classifiers was considered using foveated and unaltered images as input. The different classifiers, known collectively as the ‘what ’ module were then integrated with a ‘where ’ prediction module to determine gaze location. This module estimates object state based on previous state history, using a traditional Kalman filter. Gaze selection is then performed starting at the expected object position and moving outward using a randomized diamond shaped search pattern. The optimal gaze location is selected and used to estimate track object position. Following completion of the tracking system, its performance was evaluated using dynamic digit videos generated from the MNIST dataset. Test videos included four classes of randomly generated motion profiles and varying degrees of in-scene clutter. In general, the performance of the algorithm was robust to changes in motion profile and degree of clutter. 1.

Cited by 0

Multiple object tracking using RNNs

Khizer Amin · RECERCAT (Consorci de Serveis Universitaris de Catalunya) · 2017

Tracking multiple objects simultaneously from a single camera is popular in surveillance systems. The standard methodology to address the problem is decomposing it in the object appearance modeling, object detection/tracking and data association. The majority of the related work focuses on improving

Cited by 3Open access

Object Tracking Through Multiple Cues

Houqiang Li · 2008

In this paper a novel approach for object tracking using multiple cues is presented,based on the investigation of human eye tracking.In this approach,multiple cues with serial structure are used in tracking.First,some candidate objects are found through near-neighbor principle.Then multiple cues with different priorities select the candidates in-order.At last,the only survive candidate is fined by the modified module to get more accurate tracking result.The most salient characteristic of this approach is the principles of self-organization and self-adaptation of the changing environment during tracking.Multiple cues compete and cooperate in the system,which make the tracking has strong adaptation and vitality.Experiments show the robustness and accuracy of the tracking algorithm.

Cited by 0

Object-Tracking Using Multiple Constraints

Ismail Oner Sebe · 2002

Object-tracking is one of the most popular areas of video processing; most of the methods so far are object-dependent and concentrates only on one constraint of the object. In this report, I tried to combine multiple constraints for a faster and more robust tracking algorithm. Dynamic use of multiple constraints is the ultimate goal of the project, for the specific video sequences of basketball game.

Cited by 0

“Object Tracking Using Multiple Cameras”

Lalita Gavit, Reenal Sanghavi, Manasi Parab et al. · 2015

ABSTRACT: A single camera is not capable of covering large areas. Hence, we use multiple cameras which are placed in different sections of the area to be covered. The cameras are placed with overlapping region between field of view (FOV) of different cameras. Each camera will capture the video of its FOV. The system is intelligent enough to track people successfully in multiple perspective imagery, by establishing correspondence between objects captured in multiple cameras. Thus, it saves the tedious job of manual tracking. The methodology used to track the object is the BLOCK METHOD ANALYSIS which works on the principle of prediction. A search window for each object in the frame is acquired which helps in giving us the trajectory of the object. Continuity of this process in each frame will give the track of the object. For tracking multiple objects, the system will give labeling to every object in the frame.There can be a possibility where one object will be hidden by another object in the FOV of any one of the cameras. This problem is referred to as occlusion. In such a situation, the tracking of the hidden object should not be stopped. Hence, occlusion needs to be detected and removed. Our system deals with this problem. This project will be useful in surveillance. Need of surveillance is to monitor people or objects in areas like car parking, hotels etc for security purpose. Use of such a system will be beneficial in places which require less labour and more efficiency. Thus, it is used for public benefit.

Cited by 0