arxiv: 2511.20239 · v3 · submitted 2025-11-25 · 📡 eess.SY · cs.SY

Occlusion-Aware Multi-Object Tracking via Expected Probability of Detection

Jan Krej\v{c}\'i , Oliver Kost , Yuxuan Xia , Lennart Svensson , Ond\v{r}ej Straka This is my paper

Pith reviewed 2026-05-17 04:53 UTC · model grok-4.3

classification 📡 eess.SY cs.SY

keywords multi-object trackingocclusion handlingprobability of detectionreduced Palm densitymulti-Bernoulli mixture filtervisual trackingpoint-object sensor model

0 comments p. Extension

The pith

Multi-object tracking accounts for occlusions by assigning each object an expected probability of detection over the reduced Palm density.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes enhancing the standard point-object sensor model in multi-object tracking so that detection probability for any one object takes into account the presence of all others. It derives a principled assignment of an expected detection probability to each object, where the expectation is taken over the reduced Palm density conditional on that object's existence. This expected value therefore reflects the object's visibility to the sensor given the full set of objects. A sympathetic reader would care because the approach gives a clear and systematic way to incorporate joint visibility uncertainties into the tracking recursion rather than handling them in an ad-hoc fashion.

Core claim

By modifying the point-object model to let the probability of detection depend on the presence of every object, the method assigns each object an expected probability of detection taken over the reduced Palm density. The expectation is formed conditionally on the object's existence, so the assigned probability incorporates the object's visibility relative to the sensor under the joint configuration of all objects. Unlike existing methods, this construction accounts for uncertainties related to all objects in a clear and manageable way and is demonstrated inside the multi-Bernoulli mixture filter with marks for visual tracking.

What carries the argument

Expected probability of detection for each object, formed by taking the expectation over the reduced Palm density conditional on the object's existence.

If this is right

The expected detection probability can be inserted directly into the multi-Bernoulli mixture filter recursion for visual tracking.
Joint visibility effects are handled systematically rather than through separate heuristic corrections.
The same construction applies to any multi-object filter that maintains a representation of object existence uncertainty.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same expected-probability construction could be tested in other sensor models that admit a point-object approximation.
Crowded-scene applications such as pedestrian or vehicle tracking may see measurable gains once the reduced Palm density is approximated efficiently.
The method suggests a general pattern for folding configuration-dependent sensor effects into existence-conditioned expectations.

Load-bearing premise

The reduced Palm density can be computed or approximated in a tractable way that still correctly captures the joint visibility effects of all objects under the point-object sensor model.

What would settle it

A controlled visual tracking experiment with ground-truth object positions and known occlusion patterns in which the proposed filter's position and cardinality errors are compared against an otherwise identical filter that uses a fixed or independent detection probability for each object.

Figures

Figures reproduced from arXiv: 2511.20239 by Jan Krej\v{c}\'i, Lennart Svensson, Oliver Kost, Ond\v{r}ej Straka, Yuxuan Xia.

**Figure 1.** Figure 1: Illustration of the pedestrian state xk (52) in red. The image plane is depicted in turquoise, and C is the camera center. The SPO model [61] involves the constant PD=0.529, that was estimated from the entire MOT-17 training dataset [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗

**Figure 2.** Figure 2: PoD as a function of visibility ratio. An apparent model–data mismatch was revealed in [61] by computing an estimate PD(v) of the PoD conditionally on the visibility ratio v∈[0, 1], where v is available in ground-truth (GT) data [62, [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

read the original abstract

This paper addresses multi-object systems, where objects may occlude one another relative to the sensor. The standard point-object model for detection-based sensors is enhanced so that the probability of detection considers the presence of all objects. A principled tracking method is derived, assigning each object an expected probability of detection, where the expectation is taken over the reduced Palm density, which means conditionally on the object's existence. The assigned probability thus considers the object's visibility relative to the sensor, under the presence of other objects. Unlike existing methods, the proposed method systematically accounts for uncertainties related to all objects in a clear and manageable way. The method is demonstrated through a visual tracking application using the multi-Bernoulli mixture (MBM) filter with marks.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper uses the reduced Palm density to define an expected detection probability that accounts for joint occlusions in a multi-Bernoulli mixture filter, but the computational cost of that expectation is the open question.

read the letter

The core idea is to replace the usual fixed or state-dependent detection probability with an expectation taken over the reduced Palm density of the other objects. This gives each target a visibility probability that reflects the presence of everything else without needing separate occlusion rules. That framing is new in the MBM literature they cite and it is a clean way to fold the point-process view of the scene into the filter update. They demonstrate it on a visual tracking example, which at least shows the method can be plugged into an existing implementation. The derivation appears to stay within standard random finite set tools, so the formal steps look internally consistent on the surface. The main soft spot is tractability. Conditioning on one object via the Palm density still leaves an integral over all possible placements of the remaining points, and visibility is a deterministic geometric function of the whole configuration. Unless the paper supplies an exact factorization, a closed-form marginal, or a controlled approximation that preserves the higher-order occlusion statistics, the expectation will scale poorly with object count. The abstract does not spell out how they compute or approximate it, so that step needs checking. The experiments are limited to one visual scenario, which is reasonable for a first paper but leaves open how the method behaves under radar or lidar clutter where the geometry differs. This work is aimed at people already using MBM or other point-process filters who want a more systematic occlusion model. It is worth sending to referees who know the RFS literature so they can verify the Palm-density step and the implementation details. I would not cite it yet, but it is solid enough to deserve a full review rather than a desk reject.

Referee Report

1 major / 2 minor

Summary. The paper proposes an occlusion-aware extension to the standard point-object sensor model for multi-object tracking. It derives a per-object expected probability of detection by taking the expectation with respect to the reduced Palm density of the remaining objects (conditionally on the object's existence). This quantity is then incorporated into the multi-Bernoulli mixture (MBM) filter and demonstrated on a visual tracking application.

Significance. If the expectation step is both exact with respect to joint visibility geometry and remains computationally tractable, the approach would supply a principled, point-process-theoretic alternative to ad-hoc occlusion handling in MOT. The explicit use of the reduced Palm density is a methodological strength that could improve uncertainty propagation across all objects.

major comments (1)

[§3] §3 (Expected Probability of Detection): the central claim that the expectation over the reduced Palm density systematically accounts for all pairwise and higher-order occlusion effects requires an explicit statement of how the integral (or sum) over the joint configuration of the other points is evaluated or approximated inside the MBM filter recursion. Under a deterministic geometric visibility model the Palm conditioning produces an exponential sum; without a closed-form marginal, factorization, or provably convergent approximation that preserves the occlusion statistics, tractability for realistic object counts is not established.

minor comments (2)

[Abstract / §2] The abstract states that a derivation exists but supplies no equations; the introduction or §2 should include a short recap of the reduced Palm density definition and its relation to the standard intensity measure.
[§2 / §4] Notation for the marked point process and the sensor model should be introduced once and used consistently; several symbols appear to be redefined between the model section and the filter update.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback on our manuscript. The major comment raises an important point about computational clarity, which we address below. We have revised the manuscript to incorporate additional explanations where this strengthens the presentation.

read point-by-point responses

Referee: [§3] §3 (Expected Probability of Detection): the central claim that the expectation over the reduced Palm density systematically accounts for all pairwise and higher-order occlusion effects requires an explicit statement of how the integral (or sum) over the joint configuration of the other points is evaluated or approximated inside the MBM filter recursion. Under a deterministic geometric visibility model the Palm conditioning produces an exponential sum; without a closed-form marginal, factorization, or provably convergent approximation that preserves the occlusion statistics, tractability for realistic object counts is not established.

Authors: We agree that an explicit description of the evaluation procedure is warranted for full reproducibility. In §3 the expected probability of detection is obtained by integrating the geometric visibility function against the reduced Palm density of the remaining objects (conditioned on the existence of the object of interest). Within the MBM filter, this integral is realized by first forming the conditional multi-object density given the object's existence and then computing the expectation as a weighted sum over the finite mixture components. Each component encodes a possible joint configuration of the other objects; the weights are the normalized posterior probabilities of those components. For the deterministic visibility model the sum is formally exponential in the number of objects, but the MBM structure permits an efficient recursive evaluation that reuses the single-object densities already maintained by the filter. In the visual-tracking implementation we further employ a Monte-Carlo approximation that draws configurations from the current posterior mixture; the approximation error vanishes as the number of samples increases, thereby preserving the occlusion statistics to arbitrary accuracy. We have added a new paragraph and accompanying pseudocode in the revised §3 that spells out these steps, the scaling with object count, and the convergence argument. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation relies on external point-process concepts

full rationale

The paper defines the expected probability of detection for each object as an expectation taken with respect to the reduced Palm density (a standard concept from point process theory). This is not self-defined in terms of the output, nor is it a fitted parameter renamed as a prediction. No load-bearing self-citations, uniqueness theorems imported from the authors' prior work, or ansatzes smuggled via citation are present in the abstract or described derivation. The method is presented as a principled extension of the point-object sensor model using established Palm conditioning, making the central claim independent of its own inputs. The tractability concern raised in the skeptic note pertains to computational feasibility rather than circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities can be identified. The method relies on standard assumptions of the point-object model and the existence of a reduced Palm density for the multi-object process.

pith-pipeline@v0.9.0 · 5436 in / 1125 out tokens · 24400 ms · 2026-05-17T04:53:13.996598+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

assigning each object an expected probability of detection, where the expectation is taken over the reduced Palm density... pΘ|x(O|x) = pΞ(O∪{x})/DΞ(x)
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Tractable EPoD computation is proposed for the prior being a multi-Bernoulli mixture (MBM) with marks

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

74 extracted references · 74 canonical work pages · 2 internal anchors

[1]

B.-N. V o, M. Mallick, Y . Bar-Shalom, S. Coraluppi, R. Osborne III, R. Mahler, and B.-T. V o,Multitarget Tracking, pp. 1–15. Wiley Enc. of Electrical and Electronics Eng., 2015

work page 2015
[2]

Blackman and R

S. Blackman and R. Popoli,Design and Analysis of Modern Tracking Systems. Artech House, 1999

work page 1999
[3]

Are we ready for autonomous driving? The KITTI vision benchmark suite,

A. Geiger, P. Lenz, and R. Urtasun, “Are we ready for autonomous driving? The KITTI vision benchmark suite,” in2012 IEEE Conf. on Comp. Vision and Pattern Rec., pp. 3354–3361, 2012

work page 2012
[4]

Online multi-object tracking with δ-GLMB filter based on occlusion and identity switch handling,

M. Abbaspour and M. A. Masnadi-Shirazi, “Online multi-object tracking with δ-GLMB filter based on occlusion and identity switch handling,”Image and Vision Computing, vol. 127, p. 104553, 2022

work page 2022
[5]

On-line visual tracking with occlusion handling,

T. Rathnayake, A. Khodadadian Gostar, R. Hoseinnezhad, R. Ten- nakoon, and A. Bab-Hadiashar, “On-line visual tracking with occlusion handling,”Sensors, vol. 20, no. 3, 2020

work page 2020
[6]

Occlusion-robust online multi-object visual tracking using a GM-PHD filter with CNN-based re-identification,

N. L. Baisa, “Occlusion-robust online multi-object visual tracking using a GM-PHD filter with CNN-based re-identification,”Journal of Visual Commun. and Image Rep., vol. 80, p. 103279, 2021

work page 2021
[7]

An occlusion- aware multi-target multi-camera tracking system,

A. Specker, D. Stadler, L. Florin, and J. Beyerer, “An occlusion- aware multi-target multi-camera tracking system,” in2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pp. 4168–4177, 2021

work page 2021
[8]

Available from: https://arxiv.org/ abs/2206.14651

N. Aharon, R. Orfaig, and B.-Z. Bobrovsky, “BoT-SORT: Robust associations multi-pedestrian tracking.” arXiv:2206.14651, 2022

work page arXiv 2022
[9]

FastTracker: Real-time and accurate visual tracking,

H. Hashempoor and Y . D. Hwang, “FastTracker: Real-time and accurate visual tracking,”arXiv:2508.14370, 2025

work page arXiv 2025
[10]

ByteTrack: Multi-object tracking by associating every detection box,

Y . Zhang, P. Sun, Y . Jiang, D. Yu, F. Weng, Z. Yuan, P. Luo, W. Liu, and X. Wang, “ByteTrack: Multi-object tracking by associating every detection box,”arXiv:2110.06864, 2021

work page arXiv 2021
[11]

SparseTrack: Multi-object tracking by performing scene decomposition based on pseudo-depth,

Z. Liu, X. Wang, C. Wang, W. Liu, and X. Bai, “SparseTrack: Multi-object tracking by performing scene decomposition based on pseudo-depth,”arXiv: 2306.05238, 2023

work page arXiv 2023
[12]

Depth perspective-aware multiple object tracking,

K. G. Quach, H. Le, P. Nguyen, C. N. Duong, T. D. Bui, and K. Luu, “Depth perspective-aware multiple object tracking,” 2023. 6We use [60, Theorem 1.8] under [60, Remark 1.13] and [60, Ex- ample 1.15(f)]. For X=R n, no further assumptions would be needed assuming the reference measure is Radon [60, Remark 1.13]

work page 2023
[13]

Tracking multiple unmanned aerial vehicles through occlusion in low-altitude airspace,

S. A. Memon, H. Son, W.-G. Kim, A. M. Khan, M. Shahzad, and U. Khan, “Tracking multiple unmanned aerial vehicles through occlusion in low-altitude airspace,”Drones, vol. 7, no. 4, 2023

work page 2023
[14]

Limits of linear multitarget tracking,

D. Musicki, “Limits of linear multitarget tracking,” in2005 7th International Conference on Information Fusion, vol. 1, p. 6, 2005

work page 2005
[15]

Tracking of multiple, partially occluded humans based on static body part detection,

B. Wu and R. Nevatia, “Tracking of multiple, partially occluded humans based on static body part detection,” in2006 IEEE Conf. on Comp. Vis. and Patt. Rec., vol. 1, pp. 951–958, 2006

work page 2006
[16]

Robust multiperson tracking from a mobile platform,

A. Ess, B. Leibe, K. Schindler, and L. van Gool, “Robust multiperson tracking from a mobile platform,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 31, no. 10, pp. 1831–1846, 2009

work page 2009
[17]

Monocular visual scene understanding: Understanding multi-object traffic scenes,

C. Wojek, S. Walk, S. Roth, K. Schindler, and B. Schiele, “Monocular visual scene understanding: Understanding multi-object traffic scenes,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 35, no. 4, pp. 882–897, 2013

work page 2013
[18]

Depth assisted occlusion handling in video object tracking,

Y . Ma and Q. Chen, “Depth assisted occlusion handling in video object tracking,” inAdvances in Visual Computing, pp. 449–460, Springer Berlin Heidelberg, 2010

work page 2010
[19]

Object tracking in the presence of occlusions via a camera network,

A. O. Ercan, A. E. Gamal, and L. J. Guibas, “Object tracking in the presence of occlusions via a camera network,” in2007 6th Int. Symp. on Inf. Proc. in Sens. Net., pp. 509–518, 2007

work page 2007
[20]

A probabilistic framework for multitarget tracking with mutual occlusions,

M. Yang, Y . Liu, L. Wen, Z. You, and S. Z. Li, “A probabilistic framework for multitarget tracking with mutual occlusions,” in 2014 IEEE Conf. Comp. Vis. Patt. Rec., pp. 1298–1305, 2014

work page 2014
[21]

Negative observations for multi- ple hypothesis tracking of dynamic extended objects,

K. Wyffels and M. Campbell, “Negative observations for multi- ple hypothesis tracking of dynamic extended objects,” in2014 American Control Conference, pp. 642–647, 2014

work page 2014
[22]

Negative information for occlu- sion reasoning in dynamic extended multiobject tracking,

K. Wyffels and M. Campbell, “Negative information for occlu- sion reasoning in dynamic extended multiobject tracking,”IEEE Transactions on Robotics, vol. 31, no. 2, pp. 425–442, 2015

work page 2015
[23]

Interactive particle filter with occlusion handling for multi-target tracking,

B. Yang and R. Yang, “Interactive particle filter with occlusion handling for multi-target tracking,” in2015 12th Int. Conf. on Fuzzy Syst. and Knowledge Disc. (FSKD), pp. 1945–1949, 2015

work page 1945
[24]

Joint monocular 3d vehicle detection and tracking,

H.-N. Hu, Q.-Z. Cai, D. Wang, J. Lin, M. Sun, P. Kraehenbuehl, T. Darrell, and F. Yu, “Joint monocular 3d vehicle detection and tracking,” in2019 IEEE/CVF Intern. Conf. on Comp. Vis. (ICCV), pp. 5389–5398, 2019

work page 2019
[25]

Dynamic occlusion handling in the PHD filter for range-only tracking: Proof of concept,

S. Jovanoska, F. Govaers, R. Thom ¨a, and W. Koch, “Dynamic occlusion handling in the PHD filter for range-only tracking: Proof of concept,” in2014 Sensor Data Fusion: Trends, Solutions, Applications (SDF), pp. 1–6, 2014

work page 2014
[26]

Multiple target tracking in occlusion area with interacting object models in urban environ- ments,

J.-F. Chen, C.-C. Wang, and C.-F. Chou, “Multiple target tracking in occlusion area with interacting object models in urban environ- ments,”Robotics and Auton. Syst., vol. 103, pp. 68–82, 2018

work page 2018
[27]

An analytical formula- tion of global occlusion reasoning for multi-target tracking,

A. Andriyenko, S. Roth, and K. Schindler, “An analytical formula- tion of global occlusion reasoning for multi-target tracking,” in2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops), pp. 1839–1846, 2011

work page 2011
[28]

Online multi-object tracking with GMPHD filter and occlusion group management,

Y .-M. Song, K. Yoon, Y .-C. Yoon, K. C. Yow, and M. Jeon, “Online multi-object tracking with GMPHD filter and occlusion group management,”IEEE Access, vol. 7, pp. 165103–165121, 2019

work page 2019
[29]

R. P. S. Mahler,Statistical Multisource-Multitarget Information Fusion. Artech House, 2007

work page 2007
[30]

R. P. S. Mahler,Advances in Statistical Multisource-Multitarget Information Fusion. Artech House, 2014

work page 2014
[31]

Streit, R

R. Streit, R. B. Angle, and M. Efe,Analytic Combinatorics for Multiple Object Tracking. Springer Nature Switzerland AG, 2021

work page 2021
[32]

A labeled random finite set online multi-object tracker for video data,

D. Y . Kim, B.-N. V o, B.-T. V o, and M. Jeon, “A labeled random finite set online multi-object tracker for video data,”Pattern Recognition, vol. 90, pp. 377–389, 2019

work page 2019
[33]

Bayesian multi-target tracking with merged measurements using labelled random finite sets,

M. Beard, B.-T. V o, and B.-N. V o, “Bayesian multi-target tracking with merged measurements using labelled random finite sets,”IEEE Trans. on Signal Processing, vol. 63, no. 6, pp. 1433–1447, 2015

work page 2015
[34]

Faa di Bruno's formula for Gateaux differentials and interacting stochastic population processes

D. E. Clark and J. Houssineau, “Faa di Bruno’s formula for Gateaux differentials and interacting stochastic population pro- cesses,”arXiv:1202.0264, 2012

work page internal anchor Pith review Pith/arXiv arXiv 2012
[35]

Real-time multiple objects tracking with occlusion handling in dynamic scenes,

T. Yang, Q. Pan, J. Li, and S. Li, “Real-time multiple objects tracking with occlusion handling in dynamic scenes,” in2005 IEEE 14 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. , No. Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), vol. 1, pp. 970–975 vol. 1, 2005

work page 2005
[36]

Appearance models for occlusion handling,

A. Senior, A. Hampapur, Y .-L. Tian, L. Brown, S. Pankanti, and R. Bolle, “Appearance models for occlusion handling,”Image and Vision Computing, vol. 24, no. 11, pp. 1233–1243, 2006

work page 2006
[37]

Extended target tracking using a Gaussian-mixture PHD filter,

K. Granstrom, C. Lundquist, and O. Orguner, “Extended target tracking using a Gaussian-mixture PHD filter,”IEEE Trans. on Aerospace and Elec. Syst., vol. 48, no. 4, pp. 3268–3286, 2012

work page 2012
[38]

Measurement-wise occlusion in multi- object tracking,

M. Motro and J. Ghosh, “Measurement-wise occlusion in multi- object tracking,” in2018 21st International Conference on Infor- mation Fusion (FUSION), pp. 2384–2391, 2018

work page 2018
[39]

Dynamic- occlusion likelihood incorporation in a PHD filter based range- only tracking system,

S. Jovanoska, F. Govaers, R. Thom ¨a, and W. Koch, “Dynamic- occlusion likelihood incorporation in a PHD filter based range- only tracking system,” in2015 18th International Conference on Information Fusion (Fusion), pp. 1078–1084, 2015

work page 2015
[40]

Modeling inter-vehicle occlusion scenarios in multi-camera traffic surveillance systems,

L. Strand, J. Honer, and A. Knoll, “Modeling inter-vehicle occlusion scenarios in multi-camera traffic surveillance systems,” in2023 26th Int. Conf. on Inf. Fus. (FUSION), pp. 1–8, 2023

work page 2023
[41]

A Bayesian filter for multi-view 3D multi-object tracking with occlusion handling,

J. Ong, B.-T. V o, B.-N. V o, D. Y . Kim, and S. Nordholm, “A Bayesian filter for multi-view 3D multi-object tracking with occlusion handling,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 5, pp. 2246–2263, 2022

work page 2022
[42]

Track initialization and re-identification for 3D multi-view multi-object tracking,

L. V . Ma, T. T. D. Nguyen, B.-N. V o, H. Jang, and M. Jeon, “Track initialization and re-identification for 3D multi-view multi-object tracking,”Information Fusion, vol. 111, p. 102496, 2024

work page 2024
[43]

Visual multi-object tracking with re-identification and occlusion handling using labeled random finite sets,

L. Van Ma, T. T. D. Nguyen, C. Shim, D. Y . Kim, N. Ha, and M. Jeon, “Visual multi-object tracking with re-identification and occlusion handling using labeled random finite sets,”Pattern Rec., vol. 156, p. 110785, 2024

work page 2024
[44]

Intertarget occlusion handling in multiextended target tracking based on labeled multi-Bernoulli filter using laser range finder,

K. Dai, Y . Wang, J.-S. Hu, K. Nam, and C. Yin, “Intertarget occlusion handling in multiextended target tracking based on labeled multi-Bernoulli filter using laser range finder,”IEEE/ASME Transactions on Mechatronics, vol. 25, no. 4, pp. 1719–1728, 2020

work page 2020
[45]

Dealing with occlusions with multi targets tracking algorithms for the real road context,

L. Lamard, R. Chapuis, and J.-P. Boyer, “Dealing with occlusions with multi targets tracking algorithms for the real road context,” in2012 IEEE Intelligent Vehicles Symposium, pp. 371–376, 2012

work page 2012
[46]

CPHD filter addressing occlusions with pedestrians and vehicles tracking,

L. Lamard, R. Chapuis, and J.-P. Boyer, “CPHD filter addressing occlusions with pedestrians and vehicles tracking,” in2013 IEEE Intelligent Vehicles Symposium (IV), pp. 1125–1130, 2013

work page 2013
[47]

D. J. Daley and D. Vere-Jones,An Introduction to the Theory of Point Processes: Volume II: General Theory and Structure, 2nd edition. Springer, 2008

work page 2008
[48]

Baddeley, I

A. Baddeley, I. B ´ar´any, R. Schneider, and W. Weil,Stochastic Geometry: Lectures given at the C.I.M.E. Summer School held in Martina Franca, Italy, September 13-18, 2004. Springer-Verlag Berlin Heidleberg, 2007

work page 2004
[49]

A random finite set conjugate prior and application to multi-target tracking,

B.-T. V o and B.-N. V o, “A random finite set conjugate prior and application to multi-target tracking,” in2011 Seventh International Conference on Intelligent Sensors, Sensor Networks and Informa- tion Processing, pp. 431–436, 2011

work page 2011
[50]

Labeled random finite sets and multi- object conjugate priors,

B.-N. V o and B.-T. V o, “Labeled random finite sets and multi- object conjugate priors,”IEEE Transactions on Signal Processing, vol. 61, no. 13, pp. 3460–3475, 2013

work page 2013
[51]

Labeled random finite sets and the Bayes multi-target tracking filter,

B.-N. V o, B.-T. V o, and D. Phung, “Labeled random finite sets and the Bayes multi-target tracking filter,”IEEE Transactions on Signal Processing, vol. 62, no. 24, pp. 6554–6567, 2014

work page 2014
[52]

Integral-transform derivations of exact closed-form multitarget trackers,

R. Mahler, “Integral-transform derivations of exact closed-form multitarget trackers,” in2016 19th Int. Conf. on Inf. Fus. (FUSION), pp. 950–957, 2016

work page 2016
[53]

Gaussian implementation of the multi-Bernoulli mixture filter,

´A. F. Garc´ıa-Fern´andez, Y . Xia, K. Granstr¨om, L. Svensson, and J. L. Williams, “Gaussian implementation of the multi-Bernoulli mixture filter,” in2019 22th Int. Conf. on Information Fusion (FUSION), pp. 1–8, 2019

work page 2019
[54]

Poisson multi-Bernoulli mixture filter: Direct derivation and implementation,

´A. F. Garc´ıa-Fern´andez, J. L. Williams, K. Granstr¨om, and L. Svens- son, “Poisson multi-Bernoulli mixture filter: Direct derivation and implementation,”IEEE Transactions on Aerospace and Electronic Systems, vol. 54, no. 4, pp. 1883–1901, 2018

work page 1901
[55]

The labeled multi-Bernoulli filter,

S. Reuter, B.-T. V o, B.-N. V o, and K. Dietmayer, “The labeled multi-Bernoulli filter,”IEEE Transactions on Signal Processing, vol. 62, no. 12, pp. 3246–3260, 2014

work page 2014
[56]

Marginal multi-bernoulli filters: RFS derivation of MHT, JIPDA, and association-based MeMBer,

J. L. Williams, “Marginal multi-bernoulli filters: RFS derivation of MHT, JIPDA, and association-based MeMBer,”IEEE Trans. on Aerospace and Elec. Syst., vol. 51, no. 3, pp. 1664–1687, 2015

work page 2015
[57]

Multi-target tracking with target state dependent detection,

T. L. Song, D. Mu ˇsicki, and Y . Kim, “Multi-target tracking with target state dependent detection,” in2012 15th International Conference on Information Fusion, pp. 324–329, 2012

work page 2012
[58]

Nonmy- opic gospa-driven gaussian bernoulli sensor management,

G. Jones, ´A. F. Garc ´ıa-Fern´andez, and C. Blackman, “Nonmy- opic gospa-driven gaussian bernoulli sensor management,”IEEE Transactions on Aerospace and Electronic Systems, vol. 60, no. 6, pp. 7628–7642, 2024

work page 2024
[59]

Trajectory Poisson multi-Bernoulli filters,

´A. F. Garc´ıa-Fern´andez, L. Svensson, J. L. Williams, Y . Xia, and K. Granstr ¨om, “Trajectory Poisson multi-Bernoulli filters,”IEEE Transactions on Signal Processing, vol. 68, pp. 4933–4945, 2020

work page 2020
[60]

Heinonen,Lectures on Analysis on Metric Spaces

J. Heinonen,Lectures on Analysis on Metric Spaces. Springer Science+Bussiness Media New York, 2001

work page 2001
[61]

Model-based multi-object visual tracking: Identification and standard model limitations,

J. Krejˇc´ı, O. Kost, Y . Xia, L. Svensson, and O. Straka, “Model-based multi-object visual tracking: Identification and standard model limitations,” in2025 28th Int. Conf. on Inf. Fus. (FUSION), pp. 1– 8, 2025

work page 2025
[62]

MOT16: A Benchmark for Multi-Object Tracking

A. Milan, L. Leal-Taixe, I. Reid, S. Roth, and K. Schindler, “MOT16: A benchmark for multi-object tracking,” arXiv:1603.00831, 2016

work page internal anchor Pith review Pith/arXiv arXiv 2016
[63]

Motchallenge: A benchmark for single- camera multiple target tracking,

P. Dendorfer, A. O˘sep, A. Milan, K. Schindler, D. Cremers, I. Reid, S. Roth, and L. Leal-Taix´e, “Motchallenge: A benchmark for single- camera multiple target tracking,”International Journal of Computer Vision, vol. 129, no. 4, pp. 845–881, 2021

work page 2021
[64]

Pedestrian tracking with monocular camera using unconstrained 3D motion model,

J. Krej ˇc´ı, O. Kost, O. Straka, and J. Dun ´ık, “Pedestrian tracking with monocular camera using unconstrained 3D motion model,” in 2024 27th Int. Conf. on Inf. Fus. (FUSION), pp. 1–8, 2024

work page 2024
[65]

HOTA: A higher order metric for evaluating multi-object tracking,

J. Luiten, A. O ˘sep, P. Dendorfer, P. Torr, A. Geiger, L. Leal- Taix´e, and B. Leibe, “HOTA: A higher order metric for evaluating multi-object tracking,”International Journal of Computer Vision, vol. 129, no. 2, pp. 548–578, 2021

work page 2021
[66]

How trustworthy are performance evaluations for basic vision tasks?,

T. T. D. Nguyen, H. Rezatofighi, B.-N. V o, B.-T. V o, S. Savarese, and I. Reid, “How trustworthy are performance evaluations for basic vision tasks?,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 7, pp. 8538–8552, 2023

work page 2023
[67]

TGOSPA metric parameters selection and evaluation for visual multi-object tracking

J. Krejˇc´ı, O. Kost, O. Straka, Y . Xia, L. Svensson, and ´A. F. Garc´ıa- Fern´andez, “TGOSPA metric parameters selection and evaluation for visual multi-object tracking.” arXiv:2412.08321, 2024

work page arXiv 2024
[68]

A metric on the space of finite sets of trajectories for evaluation of multi-target tracking algorithms,

´A. F. Garc ´ıa-Fern´andez, A. S. Rahmathullah, and L. Svensson, “A metric on the space of finite sets of trajectories for evaluation of multi-target tracking algorithms,”IEEE Transactions on Signal Processing, vol. 68, pp. 3917–3928, 2020

work page 2020
[69]

Simple online and realtime tracking,

A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, “Simple online and realtime tracking,” in2016 IEEE International Conference on Image Processing (ICIP), pp. 3464–3468, 2016

work page 2016
[70]

A. S. Kechris,Classical Descriptive Set Theory, vol. 156. Springer- Verlag New York, 1995

work page 1995
[71]

Goodman, H

I. Goodman, H. T. Nguyen, and R. Mahler,Mathematics of Data Fusion. Kluwer Academic Publishers, 1997

work page 1997
[72]

D. J. Daley and D. Vere-Jones,An Introduction to the Theory of Point Processes Volume I: Elementary Theory and Methods: 1st edition. Springer, 2003

work page 2003
[73]

Multitarget Bayes filtering via first-order multi- target moments,

R. P. S. Mahler, “Multitarget Bayes filtering via first-order multi- target moments,”IEEE Transactions on Aerospace and Electronic Systems, vol. 39, pp. 1152–1178, Oct 2003

work page 2003
[74]

Koller and N

D. Koller and N. Friedman,Probabilistic Graphical Models: Principles and Techniques. Cambridge, MA, USA: MIT Press, 2009. : 15

work page 2009