Pith. sign in

REVIEW 5 major objections 5 minor 114 references

mmFlux: Crowd Flow Analytics with Commodity mmWave MIMO Radar

T0 review · 5 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read One mmWave radar maps crowd flow without tracking anyone

desk verdict A genuinely new mmWave crowd-flow topology pipeline with an honest but underpowered evaluation—worth sending to review, conditional on fixing the metrics. read the letter →

arxiv 2507.07331 v2 pith:TWRWVU4G submitted 2025-07-09 eess.SP cs.CV

Anurag Pallaprolu , Winston Hurst , Yasamin Mostofi This is my paper · ORCID
classification eess.SPcs.CV
keywords mmWaveradarcrowdflowanalyticsfieldestimationopticaldirectedgeometricgraphsplitratiocurlanddivergenceprivacy-preservingsensing
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

mmFlux claims that a single off-the-shelf mmWave radar can recover the aggregate structure of crowd motion, including where flows split, merge, turn, and gather, without ever tracking an individual person. The paper builds a pipeline that turns sparse radar point clouds into smooth 2D flow fields, then into directed geometric graphs whose edges are dominant flow currents and whose vertices are split or merge points. Across 21 real-world experiments with crowds of up to 20 people, the reconstructed graphs match ground-truth topologies in 17 of 18 structured cases, with an average one-sided Chamfer distance of 0.45 m, an edge-orientation mean absolute error of 8.8 degrees, and a flow split-ratio mean absolute error of 0.1. The practical stakes are privacy-preserving crowd analytics: flow patterns and crowd semantics can be sensed with radio reflections, not cameras, and without per-person tracking or identity.

What carries the argument

The load-bearing object is the estimated flow field $\hat{\mathbf v}(\mathbf x)$, a 2D vector field built from pairwise Lucas-Kanade optical-flow matches between consecutive radar point-cloud frames. Because raw mmWave point clouds are sparse and noisy, the field is denoised by testing whether each point's flow directions are uniformly distributed around the circle, with a Kolmogorov-Smirnov test rejecting points that are, then pruned by connected-component size and smoothed by two rounds of median filtering. A normalized version $\hat{\mathbf v}_{\mathrm{unit}}$ is skeletonized with the Zhang-Suen thinning algorithm, and the skeleton is converted into a directed geometric graph $\hat G(\hat V,\hat E)$ whose edges carry flow-split ratios computed by counting radar returns inside polyline buffers. Semantic analysis then estimates the local Jacobian $\mathbf J(\mathbf x)$ by a least-squares fit in a neighborhood and reads off divergence and curl, mapping positive and negative divergence to dispersion and gathering and positive and negative curl to counterclockwise and clockwise turning.

What would settle it

Take a structured flow configuration with a deliberately extreme split ratio, such as 80 percent of the crowd on one branch and 20 percent on a branch partially occluded by foliage, and run mmFlux; the paper already reports that this setting caused a missed edge in one of three runs on configuration C4. A decisive test would repeat that scenario across many trials and check whether the one-sided Chamfer distance stays below the pedestrian-lane width of 0.76 m or whether the thin branch is consistently lost. A second, cleaner falsifier is to increase the per-frame displacement beyond 2 m, either by faster walking or a lower radar frame rate, and observe whether the reconstructed graph breaks exactly where equation (7)'s search neighborhood no longer contains the displaced detection.

Watch

Extended reading notes

Core claim

The paper's central claim is that aggregate crowd motion carries a describable topology that a single commodity mmWave MIMO radar can reconstruct. Concretely, mmFlux estimates a dense flow field from binary occupancy maps via optical-flow-style matching, cleans it with a Kolmogorov-Smirnov test for directional consistency plus morphological pruning and median filtering, skeletonizes the resulting field, and reads a directed geometric graph off the skeleton. The authors report that this graph agrees with hand-defined ground-truth flow topologies in 17 of 18 structured-crowd experiments and that flow split ratios are estimated to within a mean absolute error of 0.1. For both structured and diffuse crowds, the Jacobian of the flow field yields curl and divergence maps that localize sharp turns, direction-reversal boundaries, dispersions, and gatherings, including in experiments where no meaningful graph exists. Together these results are offered as evidence that flow-level sensing can replace the need for fragile per-person tracking in mmWave crowd analytics.

Load-bearing premise

Everything downstream rests on the assumption that a person seen in one radar frame appears again one frame later inside a 2 m by 2 m search box, so the optical-flow step can find the true displacement; occlusion, multipath, and missed detections can break this association with no tracking or motion model to fall back on.

Editorial extensions

If this is right

  • Crowd flow analytics can be done with radio signals alone, preserving privacy while still revealing how spaces are used.
  • Flow split ratios at junctions can be estimated accurately enough to guide retail layout, venue egress, and urban planning decisions.
  • Curl and divergence maps localize semantic events such as abrupt turns, gathering points, and panic-dispersal origins without requiring a graph model.
  • Because no per-person tracking is needed, the approach should scale to denser crowds where occlusions defeat individual-association methods.
  • The framework supplies a flow prior for crowd counting, replacing the uniform-spatial-usage assumption that earlier counting methods required.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural testable extension is to treat the curl and divergence fields as time-series detectors: when the sign of divergence at a point flips, the pipeline could flag a transition from gathering to dispersal without any labeled event data.
  • If the 2 m search window is the limiting assumption, a multi-scale or coarse-to-fine variant of the optical-flow step would likely extend mmFlux to faster-moving crowds or lower radar frame rates; the paper does not test this.
  • The reported 0.45 m Chamfer distance sits inside a typical pedestrian lane width, suggesting the graph embedding is accurate enough for spatial analytics; a stricter test would measure how often reconstructed vertices fall inside the true split or merge zone rather than averaging edge distance.
  • The same flow-field representation may transfer to other sparse sensing modalities, such as automotive radar or large Wi-Fi arrays, whenever the uniform-noise assumption behind the Kolmogorov-Smirnov filter holds.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. mmFlux proposes a privacy-preserving crowd flow analytics pipeline using a single commodity mmWave MIMO radar. The pipeline converts raw FMCW data into point clouds via range-only peak detection and targeted AoA, estimates pairwise flows with Lucas-Kanade optical flow on binary occupancy maps, denoises the resulting flow field with a Kolmogorov-Smirnov directional-consistency test, morphological pruning, and median filtering, then skeletonizes the field into a directed geometric graph whose vertices are split/merge points and whose edges are dominant flow currents. Flow split ratios are estimated by counting radar returns inside edge buffers, and curl/divergence fields computed from a least-squares Jacobian estimate are used to identify semantic events such as turns, splits, merges, dispersions, and gatherings. The paper reports experimental results from 21 experiments (18 structured, plus diffuse configurations) with crowds up to 20 people in three outdoor areas, claiming 17/18 successful topology reconstructions with mean one-sided Chamfer distance 0.45 m, edge orientation MAE 8.8 degrees, split-ratio MAE 0.1, and qualitatively validated semantic detections.

Significance. If the claimed results hold, mmFlux would be a useful step toward aggregate crowd-flow analytics that avoids individual tracking and preserves privacy, using off-the-shelf radar hardware. The graph-reconstruction formulation (skeletonization plus directional validation) and the Jacobian-based semantic analysis are well-motivated and are, to my knowledge, novel in the mmWave crowd-sensing literature. The authors are to be credited for running a nontrivial real-world data collection (up to 20 people, three areas, foliage/multipath) and for transparently discussing the one failed reconstruction. At the same time, the quantitative evidence is weaker than the abstract implies: the headline metrics exclude the failed case, use a one-sided distance, and lack baseline comparisons and error bars. The semantic claims are currently qualitative. These gaps are fixable with additional analysis, and the underlying methodology appears sound enough to warrant a revision rather than rejection.

major comments (5)
  1. [Sections 5.2 and 6] The headline graph-reconstruction metrics are computed only on the 17 successfully reconstructed graphs, and the one-sided Chamfer distance d_avg averages over estimated points only. A one-sided distance does not penalize a missing ground-truth branch, so the reported 0.45 m and 8.8 degrees do not quantify completeness of the recovered topology. Because the paper's central claim is high-fidelity graph reconstruction, the evaluation should report a two-sided Chamfer distance or an equivalent precision/recall measure on all 18 structured experiments, including the C4 case with the missed upper-route edge.
  2. [Section 5.2] The edge orientation MAE is computed for the 62 edges 'present in the successfully reconstructed graphs' and there is no accounting for spurious edges or wrong connectivity. This metric is therefore not an unbiased estimate of orientation accuracy. Please supplement it with edge-level precision and recall, a graph edit distance, or a quantitative comparison of vertex positions, and report per-experiment numbers so that the variance across the 18 runs is visible.
  3. [Section 5.2 and Table 1] The split-ratio MAE of 0.1 is averaged over the same successful experiments and depends on the recovered graph; the failed C4 case is excluded and no per-vertex or per-experiment error bars are reported. Given that split-ratio estimation is said to rely on an accurate G-hat topology, the reader cannot tell how much of the 0.1 error is due to flow-field estimation versus graph mis-reconstruction. Please report the per-vertex errors, including the C4 cases, and state the effect of the failed case on the aggregate MAE.
  4. [Section 5.3] The semantic curl/divergence claims (abrupt turns, split/merge boundaries, divergence source, gathering location) are supported only by qualitative visual comparisons in Fig. 7. Since the abstract and contributions claim that these semantics are 'accurately inferred,' please provide a quantitative evaluation, such as the distance between predicted curl/divergence peaks and ground-truth event locations for each configuration, or at least a labeled localization error per experiment.
  5. [Section 3.1, Eq. (7)] The pairwise flow estimate assumes that a detection at x_w,i has a counterpart inside the 2 m x 2 m search neighborhood in the next point cloud. Under occlusion, multipath, and missed detections, which the paper acknowledges, this association can fail, and every downstream step inherits those errors. Since no flow-field-level validation or sensitivity analysis is provided, please add a diagnostic quantifying pairwise-flow coverage (e.g., fraction of points with no reliable flow) or an ablation that measures how much downstream graph quality depends on the Lucas-Kanade neighborhood size and the KS threshold.
minor comments (5)
  1. [Section 5] The opening of Section 5 states that 'all 18 experiments' show strong visual alignment, while the next paragraph and Section 6 state that one C4 experiment was only partially successful; this is internally inconsistent and should be corrected.
  2. [Eq. (8)] The conditional definition of v_TAF(x) should be written with a cases environment; the current inline 'If |psi(x)| > 0' is awkward and could be confused with an argument of the function.
  3. [Section 5.3] The choice p_th = 1 for diffuse crowds means that the KS-based denoising rejects nothing; given that p_th is described as a trade-off parameter, a brief sensitivity discussion or an alternative way to set it for diffuse flows would improve reproducibility.
  4. [Figure 9 and Section 6] The caption says 'nearly all of the flow topology is recovered, except for one edge' while the text calls the result 'partially successful'; this is fine, but the wording should be unified so readers are not left wondering whether the miss is one edge or something larger.
  5. [Section 5.2] The Chamfer distance is defined with an integral over P-hat; after discretization it should be stated as a sum over the discretized estimated graph points so that the reader can reproduce the number exactly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all pipeline outputs are validated against independently assigned experimental ground truth, and the only self-cited component (the [70] GBM front end) is a reused input, not a renamed prediction.

full rationale

The derivation chain is self-contained. The estimated flow field, topology graph, split ratios, and semantic regions are all computed from radar point clouds generated by the Sec. 2 pipeline, and every headline result is compared against independently assigned ground truth: the designed flow topologies C1–C15 and the flow split ratios that were set in advance by assigning participants to routes. No pipeline parameter is fitted to these ground-truth labels, and no reported 'prediction' is a renamed version of its own input. Eq. (7) is a standard Lucas-Kanade least-squares optical flow step, Eq. (8) is a time average, Eq. (9)–(10) form a KS test on flow angles, and Eq. (12) is a local linear fit for the Jacobian; none of these encode the target topology or semantics. The only self-citation of note is the pretrained Gradient Boosted Machine from [70] used in Sec. 2 to binarize the binary trace map ('we binarize H using an ensemble of pretrained Gradient Boosted Machines [70]'); this is a reused front-end component, not the paper's target result, and it is exercised end-to-end in the 21 experiments rather than assumed true, so it does not constitute load-bearing circularity. Two evaluation caveats are explicitly flagged for the record, but they concern the strength of the quantitative support rather than circularity. First, Sec. 5.2 reports metrics only on the successful cases: 'For the 62 edges present in the successfully reconstructed graphs, we achieve an edge orientation MAE of 8.8°,' and the one-sided Chamfer distance is computed 'for each of the 17 successfully reconstructed graphs,' so the headline numbers cannot penalize a missing branch and the single failed C4 experiment is excluded. Second, Sec. 6 claims one 'can easily determine when to use each parameter' but then defers the principled structured-vs-diffuse discrimination ('A comprehensive exploration of such distinctive factors is a future work direction'). Neither of these is a circular reduction of a predicted quantity to its input; they are limitations in evaluation and parameter selection that should be weighed in assessing the paper's claims, not in its circularity score.

Assumptions & free parameters 7 free parameters · 8 assumptions · 0 invented entities

The framework introduces no new physical entities; it combines standard radar processing, optical flow, statistical filtering, and vector calculus. The central empirical claims rely on a set of hand-chosen thresholds, a pretrained model from prior work, and domain assumptions about flow consistency and ground-truth representativeness.

free parameters (7)
  • KS p-value threshold p_th = 0.15 for structured crowds; 1 for diffuse crowds
    Controls which flow points survive the uniform-angle null hypothesis test. Changed per crowd type; p_th=1 disables filtering for diffuse flows (Sec 5.3).
  • Minimum flow component area mu_S = 2 m^2
    Morphological pruning removes connected flow regions smaller than this; tied to small pedestrian group footprint (Sec 3.3, 5.2).
  • Median filter window mu_M = 0.25 m^2
    Two rounds of median filtering smooth the flow; window approximates individual area (Sec 3.3, 5.2).
  • Lucas-Kanade neighborhood N_o = 2 m x 2 m
    Local search window for optical flow, set by maximum expected displacement between frames (Sec 5.2).
  • Vertex-trace proximity threshold Upsilon = 1 m
    Distance threshold for partitioning traces and connecting vertices during graph construction (Sec 4.1, 5.2).
  • Edge buffer width rho = 1 m
    Poly-line buffer width defining edge bounding polygons for split-ratio point counting (Sec 4.2, 5.2).
  • Jacobian neighborhood N_L = 0.25 m^2
    Least-squares Jacobian fit window, set equal to mu_M (Sec 4.3, 5.2).
assumptions (8)
  • standard math Fourier transform and FMCW mixer model correctly map chirps to range and angle spectra (Eqs. 1-6).
    Standard radar signal processing; assumed without proof.
  • domain assumption Lucas-Kanade occupancy constancy: a detection in frame w can be matched in frame w+1 by minimizing occupancy difference over a local window (Eq. 7).
    Underlies all pairwise flow estimates; no explicit tracking or data association is used.
  • domain assumption Noise-only flows have uniformly distributed angles; true flows have a consistent directional preference (Sec 3.2).
    This justifies the KS test; fails for rotating or direction-reversing flows, which is why p_th=1 is needed for diffuse crowds.
  • domain assumption Crowd flows are spatially correlated and form contiguous regions (Sec 3.3).
    Morphological pruning and median filtering rely on this to remove isolated clusters.
  • domain assumption The pretrained Gradient Boosted Machine from the authors' prior work [70] reliably binarizes the wrapped-phase bandwidth into Binary Trace Maps in the three new test areas (Sec 2).
    Point cloud generation depends on this model; its weights and training data are not provided here.
  • domain assumption Underlying flow structure is time-invariant whenever there is a flow (footnote 1).
    Time-averaging and skeletonization assume a stable topology over the 30 s window.
  • domain assumption Unit normalization of the flow field before Jacobian analysis preserves the semantic features of interest, and the Taylor expansion of dx/dt = v_unit(x) is valid locally (Sec 4.3).
    Divergence and curl are computed on the normalized field, so they measure direction geometry rather than speed changes; the effect of normalization is not quantified.
  • domain assumption Ground-truth waypoint graphs represent the true crowd flow topology (Sec 5).
    Participants move freely between waypoints; the idealized graph is treated as ground truth for Chamfer and orientation error.

how reviews work

0 comments
Cite this review

Pith. "Pith review of mmFlux: Crowd Flow Analytics with Commodity mmWave MIMO Radar." pith.science (2026). https://pith.science/paper/TWRWVU4G

@misc{pith2026250707331,
  author       = {Pith},
  title        = {Pith review of: mmFlux: Crowd Flow Analytics with Commodity mmWave MIMO Radar},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TWRWVU4G}},
  note         = {Machine review of arXiv:2507.07331}
}
read the original abstract

In this paper, we present mmFlux: a novel framework for extracting underlying crowd motion patterns and inferring crowd semantics using mmWave radar. First, our proposed signal processing pipeline combines optical flow estimation concepts from vision with novel statistical and morphological noise filtering. This approach generates high-fidelity mmWave flow fields-compact 2D vector representations of crowd motion. We then introduce a novel approach that transforms these fields into directed geometric graphs. In these graphs, edges capture dominant flow currents, vertices mark crowd splitting or merging, and flow distribution is quantified across edges. Finally, we show that analyzing the local Jacobian and computing the corresponding curl and divergence enables extraction of key crowd semantics for both structured and diffused crowds. We conduct 21 experiments on crowds of up to 20 people across 3 areas, using commodity mmWave radar. Our framework achieves high-fidelity graph reconstruction of the underlying flow structure, even for complex crowd patterns, demonstrating strong spatial alignment and precise quantitative characterization of flow split ratios. Finally, our curl and divergence analysis accurately infers key crowd semantics, e.g., abrupt turns, boundaries where flow directions shift, dispersions, and gatherings. Overall, these findings validate mmFlux, underscoring its potential for various crowd analytics applications.

Figures

Figures reproduced from arXiv: 2507.07331 by the authors.

Figure 1
Figure 1. Sample structured crowds showing two domi￾nant flows (a) intersecting (b) splitting at the red star. Sample diffuse crowds showing individuals (c) fleeing from (d) converging to red star. individuals in the crowd, reducing the effects of resolution limitations and occlusion. Building on traditional optical flow methods in the area of vision, our novel signal processing pipeline ensures robust flow field estimation b… view at source ↗
Figure 2
Figure 2. Illustration of our proposed flow field generation pipeline. We extract flows between consecutive point [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Illustration of our proposed robust graph reconstruction pipeline. To complement the extracted directed [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Experimental setup and environments: (Top) A mmWave MIMO radar transmits FMCW pulses, which [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Graph representations of various crowd flow [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Sample estimated flow topology, 𝐺ˆ(𝑉 , ˆ 𝐸ˆ), obtained from our proposed pipeline, compared with the true flow topology, 𝐺(𝑉 , 𝐸). Our results capture the spatial distribution, connectivity patterns, and flow characteristics of the ground truth, and demonstrate the hig…
Figure 7
Figure 7. Figure 7: Sample curl and divergence fields for structured (top) and diffuse (bottom) crowd flows. (a) High negative [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Motivating the need for new flow field mod [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Case of incomplete graph reconstruction on [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

114 extracted references · 78 canonical work pages

  1. [70]

    Anurag Pallaprolu, Phillip Peng, Shaan Sandhu, Winston Hurst, and Yasamin Mostofi. 2024. Crowd Analytics with a Single mmWave Radar. In Proceedings of the 30th Annual International Conference on Mobile Computing and Networking . 830–844

    work page 2024

  2. [1]

    Juliane Adrian, Armin Seyfried, and Anna Sieben. 2020. Crowds in front of bottlenecks at entrances from the perspective of physics and social psychology. Journal of the Royal Society Interface 17, 165 (2020), 20190871

    2020

  3. [2]

    Alexandre Alahi, Kratarth Goel, Vignesh Ramanathan, Alexandre Robicquet, Li Fei-Fei, and Silvio Savarese. 2016. Social LSTM: Human trajectory prediction in crowded spaces. In Proceedings of the IEEE conference on computer vision and pattern recognition . 961–971

    2016

  4. [3]

    Aldahlawi, Vahid Akbari, and Glyn Lawson

    Razan Y. Aldahlawi, Vahid Akbari, and Glyn Lawson. 2024. A sys- tematic review of methodologies for human behavior modelling and routing optimization in large-scale evacuation planning. In- ternational Journal of Disaster Risk Reduction 110 (2024), 104638. https://doi.org/10.1016/j.ijdrr.2024.104638

    work page arXiv 2024

  5. [4]

    Andrea Alfarano, Luca Maiano, Lorenzo Papa, and Irene Amerini

  6. [5]

    Saad Ali and Mubarak Shah. 2007. A lagrangian particle dynamics approach for crowd flow segmentation and stability analysis. In 2007 IEEE conference on computer vision and pattern recognition . IEEE, 1–6

    2007

  7. [6]

    Saad Ali and Mubarak Shah. 2008. Floor fields for tracking in high density crowd scenes. In Computer Vision–ECCV 2008: 10th European Conference on Computer Vision, Marseille, France, October 12-18, 2008, Proceedings, Part II 10 . Springer, 1–14

    2008

  8. [7]

    Javier Andión, José M Navarro, Gregorio López, Manuel Álvarez- Campana, and Juan C Dueñas. 2018. Smart Behavioral Analytics over a Low-Cost IoT Wi-Fi Tracking Real Deployment. Wireless Communications and Mobile Computing 2018, 1 (2018), 3136471

    2018

Show all 114 references
  1. [8]

    Tom M. Apostol. 1974. Mathematical Analysis (2nd ed.). Addison- Wesley, Reading, MA

    1974

  2. [9]

    Federico Bartoli, Giuseppe Lisanti, Lamberto Ballan, and Alberto Del Bimbo. 2018. Context-aware trajectory prediction. In 2018 24th international conference on pattern recognition (ICPR) . IEEE, 1941– 1946

    2018

  3. [10]

    Mounir Bendali-Braham, Jonathan Weber, Germain Forestier, Lhas- sane Idoumghar, and Pierre-Alain Muller. 2021. Recent trends in crowd analysis: A review. Machine Learning with Applications 4 (2021), 100023

    2021

  4. [11]

    Emily Birnbaum. 2019. Privacy activists launch campaign against airlines using facial recognition technology. The Hill (2019). https://thehill.com/policy/technology/447066-privacy-activists- launch-campaign-against-airlines-using-facial-recognition/

    2019

  5. [12]

    Hong Cai, Belal Korany, Chitra R Karanam, and Yasamin Mostofi. 2020. Teaching rf to sense without rf training measurements.Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 4, 4 (2020), 1–22

    2020

  6. [13]

    Marco Canil, Jacopo Pegoraro, Anish Shastri, Paolo Casari, and Michele Rossi. 2023. Oracle: Occlusion-resilient and self-calibrating mmwave radar network for people tracking. IEEE Sensors Journal (2023)

    2023

  7. [14]

    Mark Cardamis, Hong Jia, Hao Qian, Wenyao Chen, Yihe Yan, Oula Ghannoum, Aaron Quigley, Chun Tung Chou, and Wen Hu. 2025. Leafeon: Towards Accurate Sensing of Leaf Water Content for Pro- tected Cropping, with mmWave Radar.IEEE Internet of Things Journal (2025)

    2025

  8. [15]

    Clayson Celes, Azzedine Boukerche, and Antonio AF Loureiro. 2019. Crowd management: A new challenge for urban big data analytics. IEEE Communications Magazine 57, 4 (2019), 20–25

    2019

  9. [16]

    Pc Charankumar, B Sunil, YP Nishant, M Sumanth Vamsi, and M Hemalatha. 2024. Using Existing CCTV Networks for Crowd Man- agement and Crime Prevention using Deep Learning. In 2024 15th npj Wireless Technology, November, 2025 Pallaprolu, et al. International Conference on Comput...

    2024

  10. [17]

    Qimin Chen, Johannes Merz, Aditya Sanghi, Hooman Shayani, Ali Mahdavi-Amiri, and Hao Zhang. 2022. UNIST: unpaired neural im- plicit shape translation network. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition . 18614–18622

    2022

  11. [18]

    Wei Chen, Lichun Sui, Zhengchao Xu, and Yu Lang. 2012. Im- proved Zhang-Suen thinning algorithm in binary line drawing appli- cations. In 2012 International Conference on Systems and Informatics (ICSAI2012). IEEE, 1947–1950

    2012

  12. [19]

    Yande Chen, Yuan He, Yimiao Sun, Awais Ahmad Siddiqi, Jia Zhang, and Xiuzhen Guo. 2024. mmTAI: Biometrics-assisted Multi-person Tracking with mmWave Radar. In 2024 IEEE 30th International Con- ference on Parallel and Distributed Systems (ICPADS) . IEEE, 26–33

    2024

  13. [20]

    San Francisco Chronicle. 2024. Here’s why I am opting out of TSA facial recognition - and you should, too. San Francisco Chronicle (2024). https://www.sfchronicle.com/opinion/openforum/article/ facial-recognition-tsa-19998085.php

    2024

  14. [21]

    Marco Costa. 2010. Interpersonal distances in group walking. Journal of Nonverbal Behavior 34 (2010), 15–26

    2010

  15. [22]

    Mauro De Sanctis, Simone Di Domenico, Davide Fioravanti, Ed- uardo Ballesteros Abellán, Tommaso Rossi, and Ernestina Cianca

  16. [23]

    Florenc Demrozi, Cristian Turetta, Fabio Chiarani, Philipp H Kindt, and Graziano Pravadelli. 2021. Estimating indoor occupancy through low-cost BLE devices.IEEE Sensors Journal 21, 15 (2021), 17053–17063

    2021

  17. [24]

    Stijn Denis, Ben Bellekens, Maarten Weyn, and Rafael Berkvens. 2020. Sensing thousands of visitors using radio frequency. IEEE Systems Journal 15, 4 (2020), 5090–5093

    2020

  18. [25]

    Saandeep Depatla, Arjun Muralidharan, and Yasamin Mostofi. 2015. Occupancy estimation using only WiFi power measurements. IEEE Journal on Selected Areas in Communications 33, 7 (2015), 1381–1393

    2015

  19. [26]

    Fangqiang Ding, Zhen Luo, Peijun Zhao, and Chris Xiaoxuan Lu. 2024. milliFlow: Scene Flow Estimation on mmWave Radar Point Cloud for Human Motion Sensing. In Proceedings of the European Conference on Computer Vision (ECCV). 1–14

    2024

  20. [27]

    Fangqiang Ding, Zhijun Pan, Yimin Deng, Jianning Deng, and Chris Xiaoxuan Lu. 2022. Self-Supervised Scene Flow Estimation With 4-D Automotive Radar. IEEE Robotics and Automation Letters 7, 3 (2022), 8233–8240. https://doi.org/10.1109/LRA.2022.3187248

    2022

  21. [28]

    Muhammad Umer Farooq, Mohamad Naufal M Saad, and Sultan Daud Khan. 2022. Motion-shape-based deep learning approach for diver- gence behavior detection in high-density crowd.The Visual Computer 38, 5 (2022), 1553–1577

    2022

  22. [29]

    John Joseph Fruin. 1970. Designing for pedestrians a level of service concept. Polytechnic University

    1970

  23. [30]

    Fruin and George R

    John J. Fruin and George R. Strakosch. 1987. Pedestrian planning and design. Elevator World

    1987

  24. [31]

    Martin Gérin-Lajoie, Carol L Richards, and Bradford J McFadyen

  25. [32]

    Antonio Greco, Alessia Saggese, and Bruno Vento. 2022. A Robust and Efficient Overhead People Counting System for Retail Applications. In Image Analysis and Processing – ICIAP 2022: 21st International Conference (Lecce, Italy). Springer-Verlag, Berlin, Heidelberg, 139–150. htt...

    2022 doi

  26. [33]

    The Guardian. 2025. Is it time to ring the alarm on internet door cameras? The Guardian (2025). https: //www.theguardian.com/world/2025/feb/09/is-it-time-to-ring- the-alarm-on-internet-door-cameras

    2025

  27. [34]

    Takehiro Habara and Ryosuke Kojima. 2024. Floor-Field-Guided Neural Model for Crowd Counting. IEEE Access (2024)

    2024

  28. [35]

    Dirk Helbing, Dirk Brockmann, Thomas Chadefaux, Karsten Donnay, Ulf Blanke, Olivia Woolley-Meza, Mehdi Moussaid, Anders Johansson, Jens Krause, Sebastian Schutte, et al. 2015. Saving human lives: What complexity science and information systems can contribute. Journal of statis...

    2015

  29. [36]

    Sung Won Hong and Dong Seog Han. 2014. Performance analysis of an environmental adaptive CFAR detector. Mathematical Problems in Engineering 2014, 1 (2014), 615704

    2014

  30. [37]

    Pingfeng Hsu, Guannan Liu, Shih-Hau Fang, Hsiao-Chun Wu, and Kun Yan. 2023. Novel robust on-line indoor occupancy counting system using mmWave radar. IEEE Sensors Journal 23, 11 (2023), 12159–12170

    2023

  31. [38]

    Kaiyuan Hu, Hongjie Liao, Mingxiao Li, and Fangxin Wang. 2024. Mmcount: Stationary crowd counting system based on commodity millimeter-wave radar. In ICASSP 2024-2024 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 56–60

    2024

  32. [39]

    Chitra R Karanam, Belal Korany, and Yasamin Mostofi. 2019. Tracking from one side: Multi-person passive tracking with WiFi magnitude measurements. In Proceedings of the 18th International Conference on Information Processing in Sensor Networks . 181–192

    2019

  33. [40]

    Hao Kong, Cheng Huang, Jiadi Yu, and Xuemin Shen. 2024. A sur- vey of mmwave radar-based sensing in autonomous vehicles, smart homes and industry.IEEE Communications Surveys & Tutorials(2024)

    2024

  34. [41]

    Hao Kong, Xiangyu Xu, Jiadi Yu, Qilin Chen, Chenguang Ma, Yingy- ing Chen, Yi-Chao Chen, and Linghe Kong. 2022. m3track: mmWave- Based Multi-User 3D Posture Tracking. In Proceedings of the 20th Annual International Conference on Mobile Systems, Applications and Services. 491–503

    2022

  35. [42]

    Belal Korany and Yasamin Mostofi. 2021. Counting a stationary crowd using off-the-shelf wifi. In Proceedings of the 19th annual international conference on mobile systems, applications, and services . 202–214

    2021

  36. [43]

    Matthias Kronauge and Hermann Rohling. 2013. Fast two- dimensional CFAR procedure. IEEE Trans. Aerospace Electron. Systems 49, 3 (2013), 1817–1823

    2013

  37. [44]

    Poonnaphop Kujareanpaisal, Rujira Mayhasap, Pin Pin Tea-makorn, Pavitra Jindahra, Stefano Starita, Nagul Cooharojananone, and Kanok- wan Atchariyachanvanich. 2024. BMA for the BMA: Boosting Mo- bility Analysis for the Bangkok Metropolitan Administration via Automated Pedestria...

    2024

  38. [45]

    Tarun Kulshrestha, Divya Saxena, Rajdeep Niyogi, and Jiannong Cao

  39. [46]

    Shenglei Li and Reiko Hishiyama. 2023. An Indoor People Counting and Tracking System using mmWave sensor and sub-sensors. IFAC- PapersOnLine 56, 2 (2023), 7096–7101

    2023

  40. [47]

    Shenglei Li, Haoran Luo, Tengfei Shao, and Reiko Hishiyama. 2024. A mmWave Sensor and Camera Fusion System for Indoor Occupancy Detection and Tracking. IEICE TRANSACTIONS on Information and Systems 107, 9 (2024), 1192–1205

    2024

  41. [48]

    Pinchao Liao, Yaolin Wan, Pingbo Tang, Chunlin Wu, Yumeng Hu, and Sichun Zhang. 2019. Applying crowdsourcing techniques in urban planning: A bibliometric analysis of research and practice prospects. Cities 94 (2019), 33–43. mmFlux: Crowd Flow Analytics with Commodity mmWave MI...

    2019

  42. [49]

    Sida Ling, Jingbo Zhao, Zhaoming Lu, Xiangming Wen, and Zhe Xiao. 2024. EasyCount: Crowd Counting Based on Easy Deployment Using Commodity Wi-Fi. In 2024 IEEE Wireless Communications and Networking Conference (WCNC). IEEE, 1–6

    2024

  43. [50]

    Hankai Liu, Xiulong Liu, Xin Xie, Xinyu Tong, and Keqiu Li. 2024. PmTrack: enabling personalized mmWave-based human tracking. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 7, 4 (2024), 1–30

    2024

  44. [51]

    Ming-Yu Liu, Oncel Tuzel, Ashok Veeraraghavan, and Rama Chel- lappa. 2010. Fast directional chamfer matching. In2010 IEEE Computer Society conference on computer vision and pattern recognition . IEEE, 1696–1703

    2010

  45. [52]

    Shangqing Liu, Yanchao Zhao, Fanggang Xue, Bing Chen, and Xiang Chen. 2019. DeepCount: Crowd counting with WiFi via deep learning. arXiv preprint arXiv:1903.05316 (2019)

    2019 arXiv

  46. [53]

    Yawen Lu, Cheng Han, Qifan Wang, Heng Fan, Zhaodan Kong, Dong- fang Liu, and Yingjie Chen. 2024. Optical Flow as Spatial-Temporal Attention Learners. IEEE Transactions on Pattern Analysis and Ma- chine Intelligence (2024)

    2024

  47. [54]

    Bruce D Lucas and Takeo Kanade. 1981. An iterative image registra- tion technique with an application to stereo vision. In IJCAI’81: 7th international joint conference on Artificial intelligence, Vol. 2. 674–679

    1981

  48. [55]

    Linbo Luo, Shangwei Xie, Haiyan Yin, Chunlei Peng, and Yew-Soon Ong. 2024. Detecting and quantifying crowd-level abnormal behav- iors in crowd events. IEEE Transactions on Information Forensics and Security (2024)

    2024

  49. [56]

    R Manjappa, V Rao, R Venkatesha Prasad, and A Sinitsyn. 2019. Es- timating crowd distribution using smart bulbs. In 2019 IEEE Global Communications Conference (GLOBECOM). IEEE, 1–6

    2019

  50. [57]

    Kanti V Mardia and Peter E Jupp. 2009. Directional statistics. John Wiley & Sons

    2009

  51. [58]

    George Marsaglia, Wai Wan Tsang, and Jingbo Wang. 2003. Evaluat- ing Kolmogorov’s distribution. Journal of statistical software 8 (2003), 1–4

    2003

  52. [59]

    Frank J Massey Jr. 1951. The Kolmogorov-Smirnov test for goodness of fit. Journal of the American statistical Association 46, 253 (1951), 68–78

    1951

  53. [60]

    Ramin Mehran, Brian E Moore, and Mubarak Shah. 2010. A streakline representation of flow in crowded scenes. In Computer Vision–ECCV 2010: 11th European Conference on Computer Vision, Heraklion, Crete, Greece, September 5-11, 2010, Proceedings, Part III 11 . Springer, 439– 452

    2010

  54. [61]

    Ramin Mehran, Alexis Oyama, and Mubarak Shah. 2009. Abnormal crowd behavior detection using social force model. In 2009 IEEE conference on computer vision and pattern recognition . IEEE, 935–942

    2009

  55. [62]

    Fucheng Miao, Youxiang Huang, Zhiyi Lu, Tomoaki Ohtsuki, Guan Gui, and Hikmet Sari. 2025. Wi-Fi sensing techniques for human activity recognition: Brief survey, potential challenges, and research directions. Comput. Surveys 57, 5 (2025), 1–30

    2025

  56. [63]

    Le Monde. 2024. Let’s beware of a post-Olympic drift in the use of AI-powered video surveillance. Le Monde (2024). https://www.lemonde.fr/en/opinion/article/2024/09/26/let-s- beware-of-a-post-olympic-drift-in-the-use-of-ai-powered-video- surveillance_6727335_23.html

    2024

  57. [64]

    Nandita M Nayak, Yingying Zhu, and Amit K Roy-Chowdhury. 2013. Vector field analysis for multi-object behavior modeling. Image and Vision Computing 31, 6-7 (2013), 460–472

    2013

  58. [65]

    News.com.au. 2024. No warning: Disturbing find in Aussie tree sparks mass surveillance outrage. News.com.au (2024). https://www.news.com.au/technology/environment/no-warning- disturbing-find-in-aussie-tree-sparks-mass-surveillance- outrage/news-story/02ad245ff18c8924437549971edad206

    2024

  59. [66]

    Kien Nguyen, Minh Le, Brett Martin, Ibrahim Cil, and Clinton Fookes

  60. [67]

    OpenCV Contributors. 2024. OpenCV: Object Tracking. https: //docs.opencv.org/3.4/dc/d6b/group__video__track.html Accessed: 2025-02-17

    2024

  61. [68]

    Anurag Pallaprolu, Aaditya Prakash Kattekola, Winston Hurst, Upa- manyu Madhow, Ashutosh Sabharwal, and Yasamin Mostofi. 2024. Crowd Size Estimation for Non-Uniform Spatial Distributions with mmWave Radar. arXiv preprint arXiv:2410.09200 (2024)

    2024 arXiv

  62. [69]

    Anurag Pallaprolu, Belal Korany, and Yasamin Mostofi. 2022. Wiffract: a new foundation for RF imaging via edge tracing. In Proceedings of the 28th annual international conference on mobile computing and networking. 255–267

    2022

  63. [71]

    Artificial Intelli- gence Review 55, 7 (2022), 5707–5729

    When AI meets store layout design: a review. Artificial Intelli- gence Review 55, 7 (2022), 5707–5729

    2022

  64. [72]

    Jacopo Pegoraro and Michele Rossi. 2021. Real-time people tracking and identification from sparse mm-wave radar point-clouds. IEEE Access 9 (2021), 78504–78520

    2021

  65. [73]

    Anthony E Perry and BD Fairlie. 1975. Critical points in flow patterns. In Advances in geophysics. Vol. 18. Elsevier, 299–315

    1975

  66. [74]

    Politico. 2025. Labor’s New Front in the AI Fight. Politico (2025). https://www.politico.com/news/2025/03/04/california-labor- federation-ai-bills-00211296

    2025

  67. [75]

    Dennis C Prieve. 2000. A Course in Fluid Mechanics with Vector Field Theory. Carnegie Mellon University 4, 6 (2000), 17

    2000

  68. [76]

    Andre Pearce, J Andrew Zhang, Richard Xu, and Kai Wu. 2023. Multi- object tracking with mmwave radar: A review.Electronics 12, 2 (2023), 308

    2023

  69. [77]

    Sandeep Rao. 2017. Introduction to mmWave sensing: FMCW radars. Texas Instruments (TI) mmWave Training Series (2017), 1–11

    2017

  70. [78]

    Liyuan Ren, Alexander Yarovoy, and Francesco Fioranelli. 2023. Grouped People Counting Using mm-wave FMCW MIMO Radar. IEEE Internet of Things Journal (2023)

    2023

  71. [79]

    Hermann Rohling. 1983. Radar CFAR thresholding in clutter and multiple target situations.IEEE transactions on aerospace and electronic systems 4 (1983), 608–621

    1983

  72. [80]

    Hermann Rohling. 2011. Ordered statistic CFAR technique-an overview. In 2011 12th International Radar Symposium (IRS) . IEEE, 631–638

    2011

  73. [81]

    Lionel Rakai, Huansheng Song, ShiJie Sun, Wentao Zhang, and Yanni Yang. 2022. Data association in multiple object tracking: A survey of recent techniques. Expert systems with applications 192 (2022), 116300

    2022

  74. [82]

    Sandeep Rao. 2018. MIMO Radar Application Note. https://www.ti. com/lit/an/swra554a/swra554a.pdf Accessed: 2025-02-11

    2018

  75. [83]

    Eliane Semaan, Erika Tejedor, Rajat Kumar Kochhar, Sverker Mag- nusson, and Stefan Parkvall. 2024. 6G spectrum - enabling the future mobile life beyond 2030 . Technical Report. Ericsson

    2024

  76. [84]

    Reza Shahbazian and Irina Trubitsyna. 2023. Human sensing by using radio frequency signals: A survey on occupancy and activity detection. IEEE Access 11 (2023), 40878–40904

    2023

  77. [85]

    Shapely Community. 2024. Shapely 2.0.6 Documentation: Buffer Reference. https://shapely.readthedocs.io/en/2.0.6/reference/shapely. buffer.html Accessed: 2025-03-07

    2024

  78. [86]

    David A. Sanchez. 2012. Ordinary Differential Equations and Stability Theory: An Introduction. Dover Publications

    2012

  79. [87]

    Zichao Shen, Jose Nunez-Yanez, and Naim Dahnoun. 2024. Advanced millimeter-wave radar system for real-time multiple-human tracking and fall detection. Sensors 24, 11 (2024), 3660

    2024

  80. [88]

    Xiaoyu Shi, Zhaoyang Huang, Dasong Li, Manyuan Zhang, Ka Chun Cheung, Simon See, Hongwei Qin, Jifeng Dai, and Hongsheng Li. 2023. Flowformer++: Masked cost volume autoencoding for pretraining optical flow estimation. In Proceedings of the IEEE/CVF conference on computer vision...

    2023

  81. [89]

    Pierre Soille et al. 1999. Morphological image analysis: principles and applications. Vol. 2. Springer

    1999

  82. [90]

    Berkan Solmaz, Brian E Moore, and Mubarak Shah. 2012. Identifying behaviors in crowd scenes using stability analysis for dynamical systems. IEEE transactions on pattern analysis and machine intelligence 34, 10 (2012), 2064–2070

    2012

  83. [91]

    Aryan Sharma, Junye Li, Deepak Mishra, Gustavo Batista, and Aruna Seneviratne. 2021. Passive WiFi CSI sensing based machine learning npj Wireless Technology, November, 2025 Pallaprolu, et al. framework for COVID-Safe occupancy monitoring. In 2021 IEEE in- ternational conferenc...

    2021

  84. [92]

    Andrew G Stove. 1992. Linear FMCW radar techniques. In IEE Pro- ceedings F (Radar and Signal Processing) . IET, 343–350

    1992

  85. [93]

    Deqing Sun, Stefan Roth, and Michael J Black. 2010. Secrets of optical flow estimation and their principles. In 2010 IEEE computer society conference on computer vision and pattern recognition . IEEE, 2432– 2439

    2010

  86. [94]

    BM Suraj, T Jaiwanth, Anushka Ghei, KS Srinivas, et al . 2024. Dy- namic Crowd Management Through Potential Field-Based Simula- tions. In 2024 8th International Conference on Smart Grid and Smart Cities (ICSGSC). IEEE, 341–346

    2024

  87. [95]

    Sheng Tan, Yili Ren, Jie Yang, and Yingying Chen. 2022. Commodity WiFi sensing in ten years: Status, challenges, and opportunities. IEEE Internet of Things Journal 9, 18 (2022), 17832–17843

    2022

  88. [96]

    George Stockman and Linda G Shapiro. 2001. Computer vision. Pren- tice Hall PTR

    2001

  89. [97]

    Texas Instruments [n. d.]. DCA1000EVM Quick Start Guide . Texas Instruments. Available online at https://www.ti.com/lit/ml/spruik7/ spruik7.pdf

  90. [98]

    Godfried Toussaint. 2008. Grids, Connectivity and Contour Tracing. https://jeff.cs.mcgill.ca/~godfried/teaching/mir-reading- assignments/Chapter-2-Grids-Connectivity-Contour-Tracing.pdf Course notes for MIR Reading Assignments, McGill University

    2008

  91. [99]

    Education US Department of Health and Welfare. 1970. Skinfolds, Body Girths, Biacromial Diameter, and SelectedAnthropometric Indices of Adults. Technical Report. National Center for Health Statistics. https://www.cdc.gov/nchs/data/series/sr_11/sr11_035acc.pdf

    1970

  92. [100]

    Girish Vaidya and Marco Zuniga. 2024. Exploiting mmWave and Deep-Learning Models to Estimate People Count in Urban Scenarios. In 2024 IEEE/ACM Ninth International Conference on Internet-of-Things Design and Implementation (IoTDI). IEEE, 73–84

    2024

  93. [101]

    Texas Instruments [n. d.]. A WR2243 Evaluation Module (A WR2243BOOST) mmWave Sensing Solution. Texas Instruments. Available online at https://www.ti.com/lit/ug/spruit8d/spruit8d.pdf

  94. [102]

    Yanni Yang, Jiannong Cao, Xuefeng Liu, and Xiulong Liu. 2018. Wi- Count: Passing people counting with COTS WiFi devices. In2018 27th International Conference on Computer Communication and Networks (ICCCN). IEEE, 1–9

    2018

  95. [103]

    Radek Zavorka, Tomas Mikulasek, Josef Vychodil, Jiri Blumen- stein, Aniruddha Chandra, Hussein Hammoud, Jan Marcin Kelner, Cezary Henryk Ziółkowski, Thomas Zemen, Christoph Mecklen- bräuker, et al. 2024. Characterizing the 80 GHz Channel in Static Scenarios: Diffuse Reflection...

    2024

  96. [104]

    Mingliang Zhai, Xuezhi Xiang, Ning Lv, and Xiangdong Kong. 2021. Optical flow and scene flow estimation: A survey. Pattern Recognition 114 (2021), 107861

    2021

  97. [105]

    Li Zhang, Caom Lin, Zongmin Zhao, Dongfeng Wang, and Chong Fu

  98. [106]

    Baiju Yan, Peng Wang, Lidong Du, Xianxiang Chen, Zhen Fang, and Yirong Wu. 2023. mmGesture: Semi-supervised gesture recognition system using mmWave radar. Expert Systems with Applications 213 (2023), 119042

    2023

  99. [107]

    Jing Zhao, Yi Xu, Xiaokang Yang, and Qing Yan. 2011. Crowd insta- bility analysis using velocity-field based social force model. In 2011 Visual Communications and Image Processing (VCIP) . IEEE, 1–4

    2011

  100. [108]

    Jiang Zhu, Menghuai Xu, Ruohai Guo, Fangyong Wang, Guangying Zheng, and Fengzhong Qu. 2024. Joint Multitarget Detection and Tracking with mmWave Radar.arXiv preprint arXiv:2412.17211(2024)

    2024 arXiv

  101. [111]

    Sensors 24 (2024)

    A Crowd Movement Analysis Method Based on Radar Particle Flow. Sensors 24 (2024). https://doi.org/10.3390/s24061899

    2024 doi

  102. [112]

    Tongjie Y Zhang and Ching Y. Suen. 1984. A fast parallel algorithm for thinning digital patterns. Commun. ACM 27, 3 (1984), 236–239

    1984

  103. [2005]

    Motor control 9, 3 (2005), 242–269

    The negotiation of stationary and moving obstructions during walking: anticipatory locomotor adaptations and preservation of personal space. Motor control 9, 3 (2005), 242–269

    2005

  104. [2019]

    IEEE Transactions on Mobile Computing 19, 3 (2019), 664–679

    Real-time crowd monitoring using seamless indoor-outdoor localization. IEEE Transactions on Mobile Computing 19, 3 (2019), 664–679

    2019

  105. [2022]

    IEEE Transactions on Vehicular Technology 71, 10 (2022), 10471–10484

    Rf-based device-free counting of people waiting in line: A modular approach. IEEE Transactions on Vehicular Technology 71, 10 (2022), 10471–10484

    2022

  106. [2024]

    Computer Vision and Image Understanding (2024), 104160

    Estimating optical flow: A comprehensive review of the state of the art. Computer Vision and Image Understanding (2024), 104160

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.