Pith. sign in

REVIEW 3 major objections 4 minor 38 references

Closed-Circuit Television Data as an Emergent Data Source for Urban Rail Platform Crowding Estimation

T0 review · 3 major / 4 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 CCTV footage alone can estimate urban rail platform crowding in near real time

desk verdict Useful dataset and benchmark for CCTV-based platform crowding, but the accuracy claims rest on an unvalidated ODX ground truth and the 'MILP' calibration is really ridge regression. read the letter →

arxiv 2508.03749 v1 pith:XEYD2E3J submitted 2025-08-03 cs.CV eess.IV

Riccardo Fiorista , Awad Abdelhalim , Anson F. Stewart , Gabriel L. Pincus , Ian Thistle , Jinhua Zhao This is my paper · ORCID
classification cs.CVeess.IV
keywords computervisionCCTVurbanrailplatformcrowdingoccupancyestimationsemanticsegmentationDeepLabV3objectdetection
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

This paper aims to establish that CCTV camera images alone—without fare-card data, additional sensors, or human observers—can produce accurate, near real-time estimates of how crowded an urban rail platform is. It compares object detection (YOLOv11, RT-DETRv2, APGCC), a custom crowd-level classifier (Crowd-ViT), and semantic segmentation (DeepLabV3) on a privacy-preserving dataset of more than 43,000 frames from WMATA platforms and reports that the calibrated segmentation approach is the most accurate. Across the studied platforms the calibrated model reaches $R^2 = 0.64$ with a mean absolute error of 56.42 passengers per train-arrival event, and several platforms with favorable camera views exceed $R^2 = 0.8$ after post-processing. If correct, transit agencies could turn existing surveillance cameras into a crowding sensor network independent of other data sources. The paper's reference standard is ODX, a post-hoc trip-assignment reconstruction that the authors acknowledge is noisy at busy transfer stations.

What carries the argument

The load-bearing mechanism is the MILP-calibrated segmentation weight map. DeepLabV3 produces a binary person/background map for each frame; the map is maximum-pooled into blocks, and the calibration solves a regularized ridge regression that learns a weight per block, so that the weighted sum of pooled person-pixel activations approximates the ODX occupancy label. The learned weights concentrate near train entrances and in regions far from the camera, which is how the method accounts for depth and passenger dispersion along the platform. Supporting machinery includes ODX as the labeling source, three-frame sampling around each train's 'magic moment' of peak platform crowding, and LightGBM aggregation that fuses per-camera features into a single platform-level estimate.

What would settle it

Take a set of held-out train-arrival events at a busy transfer platform, count actual waiting passengers from the same CCTV frames by hand, and compare those counts to ODX and to the calibrated DeepLabV3 predictions. If manual counts disagree with ODX by more than the model's reported mean absolute error of about 56 passengers, then the reported accuracy is an artifact of the reference standard rather than a measure of real crowding.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a platform's occupancy can be read directly from its CCTV frames, and that the most reliable reading comes not from counting bodies or heads but from segmenting pixels into person/background and calibrating those activations with a learned spatial weight map. The calibrated DeepLabV3 pipeline achieves the best per-train-arrival occupancy estimates of all tested methods (MAE 56.42, $R^2=0.64$), improves weighted mean absolute error by 7.37% over the next-best model, and stays competitive at the 15-minute aggregate level. The paper also shows that raw segmentation pixel counts and pixel ratios are strong predictors on their own, that detection-based methods systematically undercount dense scenes, and that aggregating multiple frames and camera views reduces error substantially.

Load-bearing premise

The reference standard for 'true' occupancy is ODX, a trip-assignment algorithm that reconstructs passenger paths after the fact and that the authors say is noisy and potentially wrong at busy transfer stations; if those estimates are off, every reported accuracy number is off with them.

Editorial extensions

If this is right

  • A transit agency with CCTV and historical occupancy labels could deploy the calibrated segmentation pipeline for near real-time platform crowding estimates without installing new sensors.
  • Agencies without occupancy labels can still use pre-trained detection models such as RT-DETRv2 with SAHI, which balance accuracy and cost, and later fine-tune them with manual labels.
  • Sampling several frames around each train arrival and combining multiple camera views materially improves accuracy, so sampling strategy is as important as model choice.
  • The image-derived occupancy estimates can serve as input features for crowding prediction and operational decision support, filling a gap left by models built on indirect proxies such as fare-card data.
  • Even uncalibrated segmentation features are strong predictors, so threshold-based crowding monitoring is feasible without any calibration data.

Reading between the lines

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

  • Editorial: If validated against manual counts, the same CCTV pipeline could be turned around to audit and correct ODX itself, converting a labeled noise source into a calibration target.
  • Editorial: The learned weight maps reveal where passengers queue, so the method could double as a tool for platform design decisions such as door allocation and barrier placement, not just real-time monitoring.
  • Editorial: Because the image-derived estimates are near real-time and univariate, they are a natural feed for short-term predictive models that fuse AFC tap data with visual crowding, though the paper does not test that combination.
  • Editorial: A testable extension is whether per-camera weight maps transfer to other stations with similar camera geometry, which would remove the per-platform calibration requirement.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper evaluates whether CCTV imagery can estimate platform occupancy on WMATA rail platforms. The authors build a 43,232-image, 18-camera dataset from seven platforms, label frames with occupancy estimates from WMATA's ODX trip-assignment algorithm, and compare object detectors (YOLOv11, RT-DETRv2, APGCC with SAHI), a fine-tuned Vision Transformer for crowd-level classification, and DeepLabV3 semantic segmentation with a novel MILP-based calibration that learns camera-specific weight maps. Performance is assessed per image, per train-arrival event, and per 15-minute bin; an LGBM model fuses per-camera features. The headline result is that calibrated DeepLabV3 achieves the best mean metrics at event level (Table 2, MAE 56.42, R2 0.64) and at 15-minute level (Table 3, MAE 54.10). The authors argue that CCTV alone can provide accurate near-real-time platform crowding estimates.

Significance. If the accuracy numbers are taken at face value, the work is useful: it is a large real-world CCTV benchmark, compares modern CV methods under a common label source, and introduces an interpretable, efficient calibration for segmentation maps that could be operationalized by agencies with occupancy labels. The paper is honest about several limitations, including ODX label noise, data imbalance, and center-platform degradation. However, because every reported metric is computed against an ODX target that the authors themselves describe as containing 'unquantifiable noise' and 'potential misestimation,' the headline claim is conditional on an unvalidated ground truth. The stress-test concern about biased ODX error is not a circularity problem but a construct-validity problem: the models may be learning to predict ODX bias rather than true occupancy. The practical recommendation is appropriate only after independent validation.

major comments (3)
  1. [Ground Truth Occupancy Data; Limitations & Discussion] The evaluation's target variable is ODX-derived event-level occupancy, which the authors concede 'is not designed to provide highly accurate estimates of rail platform occupancy, and likely introduces unquantifiable noise and potential misestimation in our study, particularly at busy transfer stations.' Since ODX labels are used to train the MILP calibration, to set Crowd-ViT class thresholds, to train the LGBM fusion model, and to compute all R2, MAE, wMAE, and 95th AE values, the reported numbers measure agreement with ODX, not with true platform occupancy. The dataset is concentrated at Metro Center and Gallery Place–Chinatown, both transfer stations where ODX uncertainty is acknowledged to be highest. The authors should validate a stratified random sample of frames or train-arrival events against manual counts or another independent measurement, report the error characteristics of ODX on that sample, and show how Table 2 and Table 3 change under plausible label noise. Without this, the operational claim of 'accurate, near real-time estimation' is not supported.
  2. [Data Acquisition and Sampling; Figure 3] The same event-level ODX occupancy value is assigned to all three sampled frames at -5s, 0s, and +5s around the magic moment, even though the visual crowding changes across those frames, as Figure 3 itself notes for the +5s frame. This injects structured label noise into both the per-image evaluation and the MILP calibration, which is learned per image. The paper should either use a single frame per event, model the time offset, or validate frame-level labels on a subset. At minimum, the authors should report whether the per-image results and the learned weight maps change if only the center frame is used for calibration.
  3. [Segmentation Map Weighting; Tables 2 and 3] The MILP formulation depends on free choices such as the pooling block size (p,q), the L2 regularization weight lambda, the detection confidence threshold (30%), and the crowd-level percentile thresholds, yet no sensitivity analysis or uncertainty quantification is reported. Table 2 reports point estimates from what appears to be a single stratified split, and Figure 17 shows that calibrated DeepLabV3 achieves R2 = 0.59 on A01-1, 0.89 on A01-2, and 0.29 on C01-C; on C01-C it is not the best model. The claim that 'the calibrated DeepLabV3 segmentation approach outperforms other models across all key metrics' is an average over platforms and should be qualified with per-platform metrics, confidence intervals, or repeated split evaluation.
minor comments (4)
  1. [Crowd Level Label Generation] The crowd-level thresholds are platform-specific percentiles computed on 2023 weekday occupancy after excluding zeros, so the 'empty' class can include occupancy up to the 50th percentile on a busy platform; the paper should clarify this when interpreting the Crowd-ViT confusion matrices and the operational meaning of each class.
  2. [Performance Evaluation, Eq. (7)] The wMAE weights depend on the observed y_t; this is acceptable for an offline evaluation metric, but the text should state explicitly that the metric is not computable in real time and is intended for retrospective assessment.
  3. [Notation and captions] Equation (2) uses A_c^t without a clear definition of the indexing conventions, Equation (3) capitalizes 'Where' mid-sentence, and Figure 14/16 captions refer to 'Gallery Place Chinatown Yellow/Green line C01-1' while Table 1 uses C01-C; these should be unified.
  4. [Operational Integration, Table 4] The table lists training time as N/A for YOLOv11, RT-DETRv2, and APGCC because they are pre-trained; a table note should explicitly say this to avoid the impression that training time was not measured.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the supervised comparisons are evaluated on held-out ODX labels, and the ODX benchmark caveat is a validity concern rather than a circular step.

full rationale

The paper's derivation chain is supervised learning from CCTV-derived features to ODX-derived occupancy labels, with evaluation on held-out train-arrival events. The MILP-based ridge-regression weight maps are fit by minimizing squared error against training ODX labels, and the reported R^2, MAE, 95th AE, and wMAE values are computed on a stratified held-out test split, so the calibrated DeepLabV3 advantage is not forced by construction. Crowd-ViT class labels are derived from ODX percentile thresholds, but the classifier is trained and tested on disjoint event sets, which is a standard classification setup rather than a self-definitional loop. The LGBM aggregators are likewise trained and evaluated on separate events. The only mildly interpretive move is describing the learned segmentation weight maps as 'accounting for depth', but this is a post-hoc description of learned weights, not an equation that assumes what it derives, and it does not make the reported predictions equivalent to their inputs. The paper's self-citations, such as reference 18 for ViT-based transit prediction, are contextual and are not load-bearing for the central claim. The substantive vulnerability is that ODX estimates are an unvalidated proxy for true platform occupancy, which the Limitations section explicitly concedes; this is a benchmark-validity caveat, not circularity. No fitted parameter is renamed as a prediction, and no claimed result reduces to its input by construction.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on four unverified premises: ODX as ground truth, transferability of pretrained models, stable spatial queuing structure, and the validity of standard train/test splits. The paper uses no invented entities, but the free parameters controlling detection thresholds, MILP pooling/regularization, crowd-class boundaries, and LGBM are not fully reported, which limits reproducibility.

free parameters (5)
  • Detection confidence threshold (30%) = 0.30
    Manually set for YOLOv11 and RT-DETRv2 to reduce false positives; directly controls all detection-based counts.
  • MILP pooling block size (p,q) = not reported
    Controls resolution of the learned weight map; different sizes are shown in Figure 13 but the values used in Tables 2-3 are not stated.
  • MILP L2 regularization lambda = not reported
    Ridge penalty in the calibration objective; its value is not given.
  • Crowd-level percentile thresholds (50/75/98) = platform-specific percentiles from 2023 ODX
    Define Empty/Low/Medium/High classes for Crowd-ViT; chosen from historical occupancy distributions, not from operational standards.
  • LGBM hyperparameters = not reported
    Gradient boosting settings (e.g., number of trees, learning rate) are not specified, so the aggregation step cannot be exactly reproduced.
assumptions (4)
  • domain assumption ODX-derived platform occupancies are a valid proxy for true platform occupancy.
    All labels and evaluation targets come from WMATA's ODX algorithm; the authors note in Limitations that ODX 'likely introduces unquantifiable noise and potential misestimation', so this assumption is load-bearing and unverified.
  • domain assumption Pre-trained COCO/PASCAL models transfer adequately to CCTV platform imagery.
    YOLOv11, RT-DETRv2, APGCC, and DeepLabV3 are used without domain-specific fine-tuning; observed systematic undercounting in high-density scenes indicates imperfect transfer.
  • domain assumption Passenger queuing behavior has stable spatial structure on each platform within the training window.
    The MILP calibration assumes fixed camera views and consistent queuing locations so that learned pixel weight maps remain valid; the paper states queuing near train doors generates 'a stable visual structure'.
  • domain assumption Standard supervised train/test splits give unbiased performance estimates.
    The evaluation assumes independence of samples; for A01-2 a random image split is used, which can put frames from the same train-arrival event in both train and test sets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Closed-Circuit Television Data as an Emergent Data Source for Urban Rail Platform Crowding Estimation." pith.science (2026). https://pith.science/paper/XEYD2E3J

@misc{pith2026250803749,
  author       = {Pith},
  title        = {Pith review of: Closed-Circuit Television Data as an Emergent Data Source for Urban Rail Platform Crowding Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XEYD2E3J}},
  note         = {Machine review of arXiv:2508.03749}
}
read the original abstract

Accurately estimating urban rail platform occupancy can enhance transit agencies' ability to make informed operational decisions, thereby improving safety, operational efficiency, and customer experience, particularly in the context of crowding. However, sensing real-time crowding remains challenging and often depends on indirect proxies such as automatic fare collection data or staff observations. Recently, Closed-Circuit Television (CCTV) footage has emerged as a promising data source with the potential to yield accurate, real-time occupancy estimates. The presented study investigates this potential by comparing three state-of-the-art computer vision approaches for extracting crowd-related features from platform CCTV imagery: (a) object detection and counting using YOLOv11, RT-DETRv2, and APGCC; (b) crowd-level classification via a custom-trained Vision Transformer, Crowd-ViT; and (c) semantic segmentation using DeepLabV3. Additionally, we present a novel, highly efficient linear-optimization-based approach to extract counts from the generated segmentation maps while accounting for image object depth and, thus, for passenger dispersion along a platform. Tested on a privacy-preserving dataset created in collaboration with the Washington Metropolitan Area Transit Authority (WMATA) that encompasses more than 600 hours of video material, our results demonstrate that computer vision approaches can provide substantive value for crowd estimation. This work demonstrates that CCTV image data, independent of other data sources available to a transit agency, can enable more precise real-time crowding estimation and, eventually, timely operational responses for platform crowding mitigation.

Figures

Figures reproduced from arXiv: 2508.03749 by the authors.

Figure 2
Figure 2. These values correspond to the moment of peak platform crowding, internally referred [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 36 canonical work pages

  1. [1]

    Miller, E., G. E. Sánchez-Martínez, and N. Nassir, Estimation of Passengers Left Behind by Trains in High-Frequency Transit Service Operating Near Capacity.Transportation Research Record: Journal of the Transportation Research Board, V ol. 2672, No. 8, 2018, pp. 497–504

    work page 2018

  2. [2]

    Tirachini, A., D. A. Hensher, and J. M. Rose, Crowding in public transport systems: Ef- fects on users, operation and implications for the estimation of demand.Transportation Research Part A: Policy and Practice, V ol. 53, 2013, pp. 36–52

    work page 2013

  3. [3]

    Li, Z. and D. A. Hensher, Crowding and public transport: A review of willingness to pay evidence and its relevance in project appraisal.Transport Policy, V ol. 18, No. 6, 2011, pp. 880–887

    work page 2011

  4. [4]

    Noursalehi, P., H. N. Koutsopoulos, and J. Zhao, Predictive decision support platform and its application in crowding prediction and passenger information generation.Transporta- tion Research Part C: Emerging Technologies, V ol. 129, 2021, p. 103139

    work page 2021

  5. [5]

    Douvaras, and K

    Kopsidas, A., A. Douvaras, and K. Kepaptsoglou, Extracting Metro Passenger Flow Pre- dictors from Network’s Complex Characteristics. InComplex Networks and Their Appli- cations XI(H. Cherifi, R. N. Mantegna, L. M. Rocha, C. Cherifi, and S. Miccichè, eds.), Springer International Publishing, Cham, 2023, pp. 529–540

    work page 2023

  6. [6]

    Wang, Q., S. Wang, D. Zhuang, H. Koutsopoulos, and J. Zhao, Uncertainty quantification of spatiotemporal travel demand with probabilistic graph neural networks.IEEE Transac- tions on Intelligent Transportation Systems, V ol. 25, No. 8, 2024, pp. 8770–8781

    work page 2024

  7. [7]

    Saleh, S. A. M., S. A. Suandi, and H. Ibrahim, Recent survey on crowd density estimation and counting for visual surveillance.Engineering Applications of Artificial Intelligence, V ol. 41, 2015, pp. 103–114

    work page 2015

  8. [8]

    Kavukcuoglu, and C

    LeCun, Y ., K. Kavukcuoglu, and C. Farabet, Convolutional networks and applications in vision. InProceedings of 2010 IEEE International Symposium on Circuits and Systems, IEEE, Paris, France, 2010, pp. 253–256

    work page 2010

Show all 38 references
  1. [9]

    Divvala, R

    Redmon, J., S. Divvala, R. Girshick, and A. Farhadi, You only look once: Unified, real- time object detection. InProceedings of the IEEE conference on computer vision and pattern recognition (CVPR), 2016

    2016

  2. [10]

    Zhao, Y ., W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y . Liu, and J. Chen, DETRs beat yolos on real-time object detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition (CVPR), 2024, pp. 16965–16974

    2024

  3. [11]

    Florez, S. L., A. G. Briones, J. Pavón, R. Fuentes-Fernández, and J. M. Corchado, Mon- itoring System for Detecting Non-inclusive Situations in Smart Cities. InTrends in Sus- tainable Smart Cities and Territories(L. F. Castillo Ossa, G. Isaza, O. Cardona, O. D. Castrillón, J. ...

    2023

  4. [12]

    Goh, C. G., W. H. Lim, J. Chua, and I. Atmosukarto, Image Analytics for Train Crowd Estimation. In2018 Digital Image Computing: Techniques and Applications (DICTA), 2018, pp. 1–6

    2018

  5. [13]

    Chen, Y .-W

    Chen, I.-H., W.-T. Chen, Y .-W. Liu, M.-H. Yang, and S.-Y . Kuo,Improving Point- based Crowd Counting and Localization Based on Auxiliary Point Guidance, 2024, arXiv:2405.10589 [cs]. Fiorista et al. 25

    2024 arXiv

  6. [14]

    Cheng, Z.-Q., Q. Dai, H. Li, J. Song, X. Wu, and A. G. Hauptmann, Rethinking Spatial In- variance of Convolutional Networks for Object Counting. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 19638–19648

    2022

  7. [15]

    Albattah, R

    Khan, K., W. Albattah, R. U. Khan, A. M. Qamar, and D. Nayab, Advances and Trends in Real Time Visual Crowd Analysis.Sensors, V ol. 20, No. 18, 2020

    2020

  8. [16]

    Akyon, F. C., S. O. Altinuc, and A. Temizel, Slicing Aided Hyper Inference and Fine- tuning for Small Object Detection.2022 IEEE International Conference on Image Pro- cessing (ICIP), 2022, pp. 966–970

    2022

  9. [17]

    Beyer, A

    Dosovitskiy, A., L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. De- hghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. InInternational Con- ference on Learni...

    2021

  10. [18]

    Abdelhalim, A. and J. Zhao, Computer vision for transit travel time prediction: an end-to- end framework using roadside urban imagery.Public Transport, 2024

    2024

  11. [19]

    Papandreou, I

    Chen, L.-C., G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs.IEEE transactions on pattern analysis and machine intelligence, V ol. 40, No. 4, 2017, pp. 834–...

    2017

  12. [20]

    Fischer, and T

    Ronneberger, O., P. Fischer, and T. Brox, U-net: Convolutional networks for biomed- ical image segmentation. InMedical image computing and computer-assisted interven- tion–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part III 18,...

    2015

  13. [21]

    Wong, V . W. H. and K. H. Law, Fusion of CCTV Video and Spatial Information for Auto- mated Crowd Congestion Monitoring in Public Urban Spaces.Algorithms, V ol. 16, No. 3, 2023, p. 154

    2023

  14. [22]

    Peppa, M. V ., D. Bell, T. Komar, and W. Xiao, Urban Traffic Flow Analysis Based on Deep Learning Car Detection from CCTV Image Series.The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, V ol. XLII-4, 2018, pp. 499–506

    2018

  15. [23]

    Wang, Y ., C. Li, F. Dou, J. Dang, and Y . Wang,Assessment of passenger flow on metro platform based on video images, 2025, type: Conference Poster

    2025

  16. [24]

    Chaurasia, and J

    Jocher, G., A. Chaurasia, and J. Qiu,Ultralytics YOLO, 2023

    2023

  17. [25]

    Lv, W., Y . Zhao, Q. Chang, K. Huang, G. Wang, and Y . Liu,RT-DETRv2: Improved Base- line with Bag-of-Freebies for Real-Time Detection Transformer, 2024, arXiv:2407.17140 [cs]

    2024 arXiv

  18. [26]

    Papandreou, F

    Chen, L.-C., G. Papandreou, F. Schroff, and H. Adam,Rethinking Atrous Convolution for Semantic Image Segmentation, 2017, arXiv:1706.05587 [cs]

    2017 arXiv

  19. [27]

    Gordon, J. B., H. N. Koutsopoulos, and N. H. Wilson, Estimation of population ori- gin–interchange–destination flows on multimodal transit networks.Transportation Re- search Part C: Emerging Technologies, V ol. 90, 2018, pp. 350–365

    2018

  20. [28]

    Ultralytics,Ultralytics GitHub repository, 2025

    2025

  21. [29]

    University, P.,RT-DETR v2 R101VD model, 2025

    2025

  22. [30]

    Maire, S

    Lin, T.-Y ., M. Maire, S. Belongie, L. Bourdev, R. Girshick, J. Hays, P. Perona, D. Ra- manan, C. L. Zitnick, and P. Dollár,Microsoft COCO: Common Objects in Context, 2014, version Number: 3. Fiorista et al. 26

    2014

  23. [31]

    Zhang, Y ., D. Zhou, S. Chen, S. Gao, and Y . Ma, Single-image crowd counting via multi- column convolutional neural network. InProceedings of the IEEE conference on computer vision and pattern recognition (CVPR), 2016

    2016

  24. [32]

    Parmar, N. J., A. Vaswani, J. Uszkoreit, L. Kaiser, N. Shazeer, A. Ku, and D. Tran, Image Transformer. InInternational Conference on Machine Learning (ICML), 2018

    2018

  25. [33]

    VainF,DeepLabV3Plus-PyTorch, 2025

    2025

  26. [34]

    Everingham, M., S. M. A. Eslami, L. Van Gool, C. K. I. Williams, J. Winn, and A. Zisser- man, The Pascal Visual Object Classes Challenge: A Retrospective.International Journal of Computer Vision, V ol. 111, No. 1, 2015, pp. 98–136

    2015

  27. [35]

    Chu, and S

    Domahidi, A., E. Chu, and S. Boyd, ECOS: An SOCP solver for embedded systems. In European control conference (ECC), 2013, pp. 3071–3076

    2013

  28. [36]

    Diamond, S. and S. Boyd, CVXPY: a Python-embedded modeling language for convex optimization.Journal of Machine Learning Research, 2016

    2016

  29. [37]

    Ke, G., Q. Meng, T. Finley, T. Wang, W. Chen, W. Ma, Q. Ye, and T.-Y . Liu, LightGBM: a highly efficient gradient boosting decision tree. InAdvances in neural information process- ing systems(I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. G...

    2017

  30. [38]

    Lundberg, S. M. and S.-I. Lee, A unified approach to interpreting model predictions. In Advances in neural information processing systems(I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, eds.), Curran Associates, Inc., 2017, V ol. 30

    2017

Pith tools

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