A Stochastic Birth-and-Death Approach for Street Furniture Geolocation in Urban Environments

Evan Murphy; Marco Viola; Vladimir A. Krylov

arxiv: 2509.10310 · v1 · pith:OAYB5EW2new · submitted 2025-09-12 · 💻 cs.CV · math.OC

A Stochastic Birth-and-Death Approach for Street Furniture Geolocation in Urban Environments

Evan Murphy , Marco Viola , Vladimir A. Krylov This is my paper

Pith reviewed 2026-05-21 21:49 UTC · model grok-4.3

classification 💻 cs.CV math.OC

keywords stochastic birth-and-deathstreet furniture geolocationenergy mapsurban asset mappingGIS integrationprobabilistic optimisationpoint process inferenceinfrastructure monitoring

0 comments

The pith

A stochastic birth-and-death algorithm infers the most probable configuration of street furniture from energy maps.

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

The paper sets out to show that a probabilistic framework using energy maps can deliver precise geolocation of street furniture in dense urban settings. Encoding likelihoods in a map-based geopositioned format lets the system fold in GIS layers, road maps, and placement rules directly. A stochastic birth-and-death optimisation algorithm then searches for the asset arrangement that best fits the combined likelihood data. This would matter for city authorities and operators who need reliable maps to plan monitoring and maintenance without exhaustive manual surveys. Evaluation on a realistic simulation drawn from Dublin street-lighting records tests whether the method scales to real-world complexity.

Core claim

By representing spatial likelihoods as energy maps in geopositioned format and running a stochastic birth-and-death optimisation algorithm, the approach infers the single most probable configuration of street-furniture assets while incorporating external geospatial constraints such as GIS layers and road maps.

What carries the argument

The stochastic birth-and-death optimisation algorithm, which proposes births (adding candidate asset positions) and deaths (removing them) to explore and converge on the configuration of highest probability under the energy map.

If this is right

Localisation accuracy improves because placement constraints and road networks are enforced during optimisation rather than applied afterward.
The number of assets need not be known in advance, since births and deaths let the configuration size adjust itself.
Multiple data sources can be fused at the map level before optimisation begins, reducing the need for separate post-processing steps.
The same framework can produce updated maps whenever new imagery or GIS updates become available.

Where Pith is reading between the lines

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

The same energy-map plus birth-and-death structure could locate other fixed urban objects whose counts are also unknown in advance.
Errors in the initial energy maps would directly limit final accuracy, suggesting that improving map construction is a natural next bottleneck to address.
Because the optimisation is stochastic, repeated runs on identical inputs could be used to estimate configuration uncertainty for risk-aware maintenance planning.

Load-bearing premise

Energy maps stored in map-based geopositioned format let the optimisation process integrate external geospatial information such as GIS layers and placement constraints without loss of accuracy or compatibility.

What would settle it

Applying the algorithm to the Dublin city-centre street-lighting simulation and measuring that the recovered asset configurations deviate substantially from the known ground-truth positions when external constraints are included.

Figures

Figures reproduced from arXiv: 2509.10310 by Evan Murphy, Marco Viola, Vladimir A. Krylov.

**Figure 2.** Figure 2: Example of detections of street lights (blue) by SBD (red) and MRF (green), guided by pairwise in [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Figure 3: Histograms of within cluster distances for noise levels [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

**Figure 4.** Figure 4: Plot of precision v. recall (left) and F1-score vs. detection distance (right) for a single simulation of [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗

read the original abstract

In this paper we address the problem of precise geolocation of street furniture in complex urban environments, which is a critical task for effective monitoring and maintenance of public infrastructure by local authorities and private stakeholders. To this end, we propose a probabilistic framework based on energy maps that encode the spatial likelihood of object locations. Representing the energy in a map-based geopositioned format allows the optimisation process to seamlessly integrate external geospatial information, such as GIS layers, road maps, or placement constraints, which improves contextual awareness and localisation accuracy. A stochastic birth-and-death optimisation algorithm is introduced to infer the most probable configuration of assets. We evaluate our approach using a realistic simulation informed by a geolocated dataset of street lighting infrastructure in Dublin city centre, demonstrating its potential for scalable and accurate urban asset mapping. The implementation of the algorithm will be made available in the GitHub repository https://github.com/EMurphy0108/SBD_Street_Furniture.

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

This paper applies birth-and-death optimization to geolocating street furniture from energy maps that fold in GIS constraints, but the abstract gives almost no technical detail on how the process is defined or validated.

read the letter

The main takeaway is that this paper uses a stochastic birth-and-death algorithm to find the best configuration of street furniture from energy maps in cities. It is mostly an application of known techniques to a new domain. The paper does well in setting up the problem for practical use by local authorities. The energy map approach in geopositioned format lets them bring in GIS layers and constraints easily, which should help with accuracy in complex urban areas. They back it with a simulation using real data from Dublin street lights, and they are making the code available, which helps others try it out. On the downside, the description stays very high level. No equations or details on how the energy is computed or how the birth and death steps work with the constraints. Without that, it is tough to see if the optimization really finds the most probable setup or if it just avoids obvious errors. The evaluation is simulation only, so we do not know how it handles real photos or noise in the energy maps. This kind of paper is aimed at researchers and practitioners in urban computing or computer vision for infrastructure. A reader who wants ideas for integrating geospatial data into point process models could find it worthwhile. I think it should go to peer review. The core idea has merit for reducing survey costs, and referees can push for the technical details and real-world tests that are missing here.

Referee Report

2 major / 2 minor

Summary. The paper proposes a probabilistic framework for geolocating street furniture in urban environments. Energy maps in geopositioned format encode spatial likelihoods and enable integration of GIS layers, road maps, and placement constraints. A stochastic birth-and-death optimization algorithm is introduced to recover the most probable asset configuration. The method is evaluated on a simulation derived from a geolocated street-lighting dataset in Dublin city centre, with code promised on GitHub.

Significance. If the birth-and-death process can be shown to converge to the MAP configuration of the posterior induced by the energy maps while respecting hard geospatial constraints, the framework would supply a flexible mechanism for incorporating external GIS data into urban asset mapping, offering a scalable alternative to purely data-driven or manual approaches.

major comments (2)

[Proposed method] The manuscript supplies no explicit functional form for the energy function that defines the target distribution, nor any statement of the birth and death intensity functions or the Metropolis-Hastings acceptance ratio. Without these, it is impossible to verify that the stochastic process targets the claimed posterior or satisfies detailed balance (see the paragraph introducing the stochastic birth-and-death optimisation algorithm).
[Energy maps and geospatial integration] The claim that map-based energy representation permits seamless integration of GIS layers and hard placement constraints is unsupported by any description of how constraints enter the proposal distribution or the energy (rejection sampling, modified intensities, or soft penalties). This detail is load-bearing for the assertion that sampled configurations remain valid under urban road-boundary rules (see the paragraph on energy-map representation).

minor comments (2)

[Experiments] The simulation-based evaluation is described only at a high level; no numerical error metrics, convergence diagnostics, ablation on constraint handling, or comparison against baseline geolocation methods are reported.
[Implementation] The GitHub repository link is given but the manuscript contains no description of the code structure, input data format, or instructions for reproducing the Dublin simulation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review. The comments highlight important areas where the manuscript requires greater mathematical precision to substantiate the claims. We address each point below and will revise the manuscript accordingly.

read point-by-point responses

Referee: [Proposed method] The manuscript supplies no explicit functional form for the energy function that defines the target distribution, nor any statement of the birth and death intensity functions or the Metropolis-Hastings acceptance ratio. Without these, it is impossible to verify that the stochastic process targets the claimed posterior or satisfies detailed balance (see the paragraph introducing the stochastic birth-and-death optimisation algorithm).

Authors: We agree that the explicit functional forms are necessary for rigorous verification. The original manuscript described the energy function and birth-and-death process at a conceptual level to emphasize the overall framework. In the revision we will insert the precise mathematical definitions: the energy function E(x) = -log p(x|data, constraints), the birth intensity λ_b(x) and death intensity λ_d(x) derived from the energy map, and the Metropolis-Hastings acceptance ratio that ensures detailed balance with respect to the target posterior. These additions will confirm convergence to the MAP configuration while respecting the hard geospatial constraints. revision: yes
Referee: [Energy maps and geospatial integration] The claim that map-based energy representation permits seamless integration of GIS layers and hard placement constraints is unsupported by any description of how constraints enter the proposal distribution or the energy (rejection sampling, modified intensities, or soft penalties). This detail is load-bearing for the assertion that sampled configurations remain valid under urban road-boundary rules (see the paragraph on energy-map representation).

Authors: We acknowledge that the mechanism for enforcing hard constraints was described only at a high level. In the revised manuscript we will explicitly state that road-boundary and placement constraints are incorporated by (i) modulating the birth intensity function to zero outside permitted regions and (ii) applying rejection sampling on proposals that violate hard GIS rules. This ensures all sampled configurations remain valid without relying on soft penalties alone, directly supporting the claim of seamless geospatial integration. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation remains self-contained

full rationale

The paper presents a probabilistic framework using energy maps in geopositioned format and introduces a stochastic birth-and-death optimisation algorithm to infer the most probable asset configuration. No explicit equations, energy functions, birth/death intensity functions, or fitted parameters are quoted or described in the provided abstract and text excerpts. No self-definitional reductions, fitted inputs renamed as predictions, or load-bearing self-citations appear. The evaluation on a Dublin simulation dataset is presented as external validation rather than a closed loop. The derivation chain is therefore independent of its inputs by construction and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review performed on abstract only; therefore the ledger records only the modeling assumptions explicitly named in the abstract. No free parameters or invented entities are described.

axioms (1)

domain assumption Energy maps encode the spatial likelihood of object locations and can be represented in geopositioned format to integrate GIS layers and placement constraints.
Stated as the basis for the probabilistic framework in the second paragraph of the abstract.

pith-pipeline@v0.9.0 · 5696 in / 1220 out tokens · 54824 ms · 2026-05-21T21:49:40.749645+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/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

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

A stochastic birth-and-death optimisation algorithm is introduced to infer the most probable configuration of assets... energy function H: Γ(V)→R... unary term U(x,r) and pairwise interaction term I(x,r;g)
IndisputableMonolith/Foundation/BranchSelection.lean branch_selection unclear

?

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

PGibbs(g;β)=1/Z e^{-β H(g)} ... as β→∞ ... converge weakly to the distribution of global minimisers

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

7 extracted references · 7 canonical work pages

[1]

Descombes, X., Minlos, R., and Zhizhina, E. (2009). Object extraction using a stochastic birth-and-death dynamics in continuum. Journal of Mathematical Imaging and Vision , 33(3):347--359

work page 2009
[2]

Data.gov.ie, dublin city council, public lighting

Dublin City Council (2021). Data.gov.ie, dublin city council, public lighting. https://data.gov.ie/dataset/street-lighting-dublin-city. Accessed: May 2025

work page 2021
[3]

A., Kenny, E., and Dahyot, R

Krylov, V. A., Kenny, E., and Dahyot, R. (2018). Automatic discovery and geotagging of objects from street view imagery. Remote Sensing , 10(5):661

work page 2018
[4]

D., Produit, T., and Nassar, A

Lefevre, S., Tuia, D., Wegner, J. D., Produit, T., and Nassar, A. S. (2017). Toward seamless multiview scene analysis from satellite to street level. Proceedings of the IEEE , 105(10):1884--1899

work page 2017
[5]

J., Ulicny, M., Manzke, M., and Dahyot, R

Liu, C. J., Ulicny, M., Manzke, M., and Dahyot, R. (2021). Context aware object geotagging. In Irish Machine Vision and Image Processing Conference , pages 65--72

work page 2021
[6]

Mabon, J., Ortner, M., and Zerubia, J. (2024). Learning point processes and convolutional neural networks for object detection in satellite images. Remote Sensing , 16(6):1019

work page 2024
[7]

Sun, Y., Ye, Y., Kang, J., Fernandez-Beltran, R., Feng, S., Li, X., Luo, C., Zhang, P., and Plaza, A. (2023). Cross-view object geo-localization in a local region with satellite imagery. IEEE Transactions on Geoscience and Remote Sensing , 61:1--16

work page 2023

[1] [1]

Descombes, X., Minlos, R., and Zhizhina, E. (2009). Object extraction using a stochastic birth-and-death dynamics in continuum. Journal of Mathematical Imaging and Vision , 33(3):347--359

work page 2009

[2] [2]

Data.gov.ie, dublin city council, public lighting

Dublin City Council (2021). Data.gov.ie, dublin city council, public lighting. https://data.gov.ie/dataset/street-lighting-dublin-city. Accessed: May 2025

work page 2021

[3] [3]

A., Kenny, E., and Dahyot, R

Krylov, V. A., Kenny, E., and Dahyot, R. (2018). Automatic discovery and geotagging of objects from street view imagery. Remote Sensing , 10(5):661

work page 2018

[4] [4]

D., Produit, T., and Nassar, A

Lefevre, S., Tuia, D., Wegner, J. D., Produit, T., and Nassar, A. S. (2017). Toward seamless multiview scene analysis from satellite to street level. Proceedings of the IEEE , 105(10):1884--1899

work page 2017

[5] [5]

J., Ulicny, M., Manzke, M., and Dahyot, R

Liu, C. J., Ulicny, M., Manzke, M., and Dahyot, R. (2021). Context aware object geotagging. In Irish Machine Vision and Image Processing Conference , pages 65--72

work page 2021

[6] [6]

Mabon, J., Ortner, M., and Zerubia, J. (2024). Learning point processes and convolutional neural networks for object detection in satellite images. Remote Sensing , 16(6):1019

work page 2024

[7] [7]

Sun, Y., Ye, Y., Kang, J., Fernandez-Beltran, R., Feng, S., Li, X., Luo, C., Zhang, P., and Plaza, A. (2023). Cross-view object geo-localization in a local region with satellite imagery. IEEE Transactions on Geoscience and Remote Sensing , 61:1--16

work page 2023