Pith. sign in

REVIEW 3 major objections 6 minor 55 references

Predicting Internet Connectivity in Schools: A Feasibility Study Leveraging Multi-modal Data and Location Encoders in Low-Resource Settings

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read School internet access can be predicted from free satellite and survey data with 80% accuracy in Botswana and 73% in Rwanda, the paper claims.

desk verdict Useful open dataset and honest feasibility study, but spatial leakage from overlapping buffers and test-set buffer selection inflate the headline accuracies. read the letter →

arxiv 2412.12188 v1 pith:WFQ6KBXE submitted 2024-12-13 eess.IV cs.AIcs.CVcs.SI

Kelsey Doerksen , Casper Fibaek , Rochelle Schneider , Do-Hyung Kim , Isabelle Tingzon This is my paper · ORCID
classification eess.IVcs.AIcs.CVcs.SI
keywords schoolinternetconnectivityEarthObservationmachinelearninglocationencoderssatelliteimagerydigitaldivideBotswanaRwanda
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 feasibility study tries to establish that machine learning can predict whether a school has internet access using only free, openly available satellite imagery and survey data, at accuracies that are useful to governments planning digital infrastructure. Its best models combine satellite-derived features (land cover, night-time lights, settlement type, population, human modification, distance to power lines, and speed-test data) with auxiliary school records, reaching 80% accuracy and 0.81 F1 in Botswana and 73% accuracy and 0.74 F1 in Rwanda. The paper also reports that hand-crafted satellite features outperform all four tested location-encoder embeddings on this task, although adding embeddings helped slightly in Botswana. The authors present this as a practical, low-cost screening approach for low-resource settings and release cleaned, labelled datasets so other countries can replicate the pipeline.

What carries the argument

The load-bearing object is the engineered feature vector: for each school, summary statistics such as mean, variance, mode, max, min, and class percentages are computed inside a 1,000-metre satellite-imagery buffer around the school's coordinates, drawn from MODIS land cover, VIIRS night-time lights, the Global Human Settlement Layer, Gridded Population of the World, and Global Human Modification, together with distance to the nearest power-grid transmission line and Ookla mobile and fixed speed-test statistics. A location encoder is a neural network that turns a latitude/longitude coordinate into a vector embedding summarising the surrounding area from satellite imagery; the paper compares SatCLIP, GeoCLIP, CSP, and the new ESA PhilEO VHR Pre-cursor model as alternative or additive feature sources. Shallow classifiers (random forest, gradient boosting, SVM, logistic regression, XGBoost, and MLP) map these features to a binary connected/not-connected label. The decisive extension is auxiliary school data, which is what pushes accuracy up and false positives down.

What would settle it

Re-run the same experiments with a spatial split that keeps all schools within 1,000 metres of each other in the same fold; if accuracy and F1 in Botswana and Rwanda drop substantially, or false-positive rates rise, relative to the reported 80%/0.81 and 73%/0.74, then the random split is the reason. A complementary check would be to field-validate predictions in a fresh district that contributed no schools to training.

Watch

Extended reading notes

Core claim

The central claim is that school internet connectivity can be predicted from multi-modal open geospatial data with practically useful accuracy in low-resource countries. Using a 1,000-metre satellite buffer around each school to compute features from MODIS land cover, VIIRS night-time lights, the Global Human Settlement Layer, gridded population, and global human modification, plus distance to power-grid lines and Ookla speed-test statistics, the best shallow classifiers outperform all four location-encoder embedding spaces on their own. Adding auxiliary school-level information (education level, distance to mobile network towers, one-hot administrative region, and school-age population counts) improves results further: accuracy and F1 reach 80% and 0.81 in Botswana and 73% and 0.74 in Rwanda, while false-positive rates drop from 48% to 19% in Botswana and 64% to 26% in Rwanda. The paper claims this demonstrates a viable, low-cost alternative to exhaustive school surveys, while its Kigali case study shows that the model struggles in dense urban areas where connected and unconnected schools sit within a few hundred metres of each other.

Load-bearing premise

The paper's results assume that a random 70/15/15 split of schools into training, validation, and test sets is a fair test, even though the 1,000-metre satellite windows around nearby schools overlap, so the model may have effectively seen parts of a test school's surroundings during training.

Editorial extensions

If this is right

  • Governments and development agencies can produce national school-connectivity maps from free satellite and survey data in countries that cannot afford exhaustive ground surveys, and use those maps to prioritise infrastructure investment.
  • Adding basic administrative and school-record information to satellite features is the main lever for reducing false positives, the error that would waste resources by treating unconnected schools as connected.
  • The public release of cleaned school-connectivity labels and the feature-generation pipeline enables replication of the study in other low-resource countries without proprietary data.
  • Generic location-encoder embeddings alone are not a sufficient substitute for domain-engineered features in this task, suggesting that pretrained geographic foundation models need fine-tuning or higher-resolution inputs for connectivity prediction.
  • In dense urban settings, predictions for schools less than a few hundred metres apart are unreliable, so operational deployment would need to combine model outputs with local verification.

Reading between the lines

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

  • Editorial extension: The two-country result does not establish global generalization; the same pipeline should be tested in a geographically diverse set of low-resource countries, and the released dataset makes such a test straightforward.
  • Editorial extension: Because the 1,000-metre satellite windows around nearby schools overlap, the random 70/15/15 split may leak spatial context into the test set; a cluster-based or buffer-excluded split would be a sharper test of the reported 80% and 73% accuracy figures.
  • Editorial extension: The false-positive reduction from auxiliary data points toward a practical product: a prioritisation score combining satellite context with school records, which governments could update as new schools are connected.
  • Editorial extension: The underperformance of location encoders here suggests that for infrastructure tasks, frozen generic embeddings should be evaluated against simple engineered features before being adopted, rather than assumed to carry the needed socio-economic signal.
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 / 6 minor

Summary. The paper investigates whether machine learning with freely available Earth Observation (EO) data and ground-based auxiliary data can predict binary internet connectivity status of schools in Botswana and Rwanda. The authors construct engineered features from a 1,000m buffer around each school (MODIS, VIIRS nightlight, GHSL, population, distance to transmission lines, Ookla speedtest data), extract embeddings from four location-encoder models (SatCLIP, GeoCLIP, CSP, and their PhilEO VHR), and train six shallow classifiers. They report that engineered features outperform location-encoder embeddings, and that adding auxiliary school information improves performance, achieving 80% accuracy and 0.81 F1 in Botswana and 73% and 0.74 in Rwanda. They also present a Kigali case study and a feature importance analysis.

Significance. If the reported accuracies are reliable, the work would demonstrate a low-cost, open-data screening tool for school connectivity in low-resource settings, with potential operational value for Giga/Project Connect. The paper's strengths include the release of code and preprocessed data, the systematic comparison of four location encoders across six classifiers, reporting of variance over five runs for many experiments, and a candid discussion of limitations including label staleness and buffer overlap. However, the evaluation design currently precludes the absolute accuracy claims from being taken at face value, because the random school-level split does not separate spatially overlapping feature buffers and the 1,000m buffer radius was selected using the test set.

major comments (3)
  1. [Section 3.1, Section 4, Section 5, Appendix C] The evaluation premise is undermined by spatial leakage from overlapping 1,000m buffers. The paper's own Kigali case study (Section 5) notes schools less than 400m apart with different connectivity labels, and Appendix C shows overlapping buffers. With a random 70/15/15 split (Section 4), a test school can have training schools within 2,000m, so their engineered features (MODIS, VIIRS, GHSL, population, and distance-to-grid variables) are nearly identical. The model can therefore memorize local feature patterns rather than learn a generalizable connectivity signal, inflating the test accuracy, F1, and the apparent benefit of auxiliary data in Table 5. This is a load-bearing issue for the central feasibility claim, and should be addressed by spatial cross-validation (e.g., grouping schools into spatially disjoint clusters or excluding training samples within the buffer radius) and by reporting the degree of feature overlap between train and test sets.
  2. [Section 3.1, Appendix C, Table 7] The 1,000m buffer radius was selected based on test-set performance. Section 3.1 states that the buffer extent was chosen because '1,000m yielded the best results for test set F1 score and accuracy,' and Appendix C confirms that the comparative analysis used the test set to select this hyperparameter. Using the same test set for both model selection and final evaluation makes the reported accuracies optimistically biased. The buffer radius should be selected on a validation set or via nested cross-validation, and the final test numbers should be reported only after that selection.
  3. [Section 6, Table 5] Label staleness is acknowledged but not quantified. The paper states that 'there is an undefined latency between connection and updated record,' meaning ground-truth labels may not reflect true current connectivity. Because the headline claims are absolute accuracy and F1 values (80%/0.81 in Botswana and 73%/0.74 in Rwanda, Table 5), this labeling noise—especially if it is spatially correlated—directly weakens the conclusion that the model predicts true connectivity. The paper should either quantify the expected label noise (e.g., by comparing against a sample of ground-validated schools) or temper the claims to 'predicting the current Project Connect database labels' rather than actual school connectivity.
minor comments (6)
  1. [Abstract, Section 2, Section 3.2, Section 5] There are several typographical errors: 'digital literary' should be 'digital literacy' in the abstract, 'feasability' should be 'feasibility' in Section 2, 'datatset' should be 'dataset' in Section 3.2, and 'construing' in Section 5 appears to be a typo for 'preventing' or 'confusing.'
  2. [References] The reference 'N. atledge' should be 'N. Ratledge' (Nature paper 'Using machine learning to assess the livelihood impact of electricity access').
  3. [Table 5] Table 5 reports no variance or confidence intervals for the auxiliary-data results, whereas the appendix (Tables 13-24) provides variance over five runs for the other feature spaces; the same variance reporting should be added for Table 5 to support the reliability of the headline numbers.
  4. [Table 3] In Table 3 (BW A), the SVM with SatCLIP-ViT16-L40 achieves accuracy 0.19, far below the majority-class baseline; this anomaly is not discussed and suggests a degenerate hyperparameter configuration, and it should be either explained or excluded from the comparison.
  5. [Table 5] The false-positive rate (FP) in Table 5 is not explicitly defined; the paper should state its denominator (e.g., FP/(FP+TN)) and how it is averaged across folds or runs.
  6. [Section 5, Figure 4] The feature importance analysis is based on a single model (Random Forest); the paper should note that these importances may not be stable across classifiers and should avoid over-generalizing from one model's ranking.

Circularity Check

1 steps flagged · score 4.0 of 10

The buffer radius is selected on the test set, and the same test set then supports the headline accuracies, so part of the reported prediction is fitted to the test data; the feature-space comparison itself is not circular.

  1. fitted input called prediction [Appendix C, 'Buffer extent selection' (Table 7 and Figure 7 discussion)]
    "Table 7 and Figure 7 summarize the average test set F1 and accuracy scores over varying buffer sizes for all classifiers. ... From these results, we select the 1000m buffer model's results to highlight in the paper and iterate with additional experimentation incorporating auxiliary school information."

    The 1,000m buffer is chosen by comparing test-set F1 and accuracy across five radii, and the paper then reports final test-set accuracies (80%/0.81 in Botswana and 73%/0.74 in Rwanda) from models using that chosen buffer. The headline test metrics are therefore the same objective used to select the buffer, so they are not an unbiased prediction of held-out performance; the buffer radius is a fitted input that is subsequently treated as part of the model's reported predictive result. This is a partial, not total, circularity because the central comparison among engineered features, embeddings, and auxiliary data still involves genuine supervised learning on independent labels.

full rationale

The paper is an empirical supervised-classification study, so most of its derivation chain is not circular: features, embeddings, and labels are distinct, and the models are evaluated on a held-out split. The PhilEO VHR model is self-cited (Fibaek et al. 2024) but is used only as a frozen feature extractor and is assessed on held-out labels, so that citation does not smuggle in the target result. The one concrete circular step is the buffer-extent selection in Appendix C: the authors choose 1,000m by maximizing test-set F1/accuracy and then use that same test set to report the headline 80%/0.81 and 73%/0.74 numbers, making those numbers a fitted-input-called-prediction result. A separate spatial-leakage concern (1,000m buffers of nearby train and test schools overlap, as acknowledged in Section 5 and Appendix C) is a real threat to the validity of the reported accuracies, but it is evaluation leakage rather than a derivation-equivalence, so it does not count as circularity under the definitions used here. Overall the central claim is not forced by definition or by a self-citation chain, so the score is moderate.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

No physical entities are invented. The ledger captures the data-dependent modeling choices and domain assumptions the results rely on; the main free parameter is the test-selected buffer radius, which touches the reported accuracy directly.

free parameters (1)
  • Feature buffer radius = 1000m
    Selected by comparing test set F1 and accuracy across 300m, 500m, 750m, 1000m and 5000m buffers (Appendix C). Choosing a modeling parameter on the test set makes the final reported metrics optimistic.
assumptions (4)
  • domain assumption Project Connect ground truth labels reflect true school connectivity
    Section 3.1 uses government-reported labels; Section 6 acknowledges undefined latency between connection and record update, so labels may be stale.
  • domain assumption Satellite-derived features and proximity to power grid and speedtest activity are predictive of school connectivity
    Section 3.1 justifies features through poverty-mapping literature; no causal mechanism is established.
  • domain assumption Random 70/15/15 split with overlapping 1000m buffers does not leak information between train and test
    Section 4 describes the split; Appendix C and the Kigali case study show nearby schools with overlapping buffers, but no spatial blocking is applied.
  • domain assumption Pre-trained location encoders provide transferable embeddings for this downstream task
    Section 3.2 and Appendix E rely on SatCLIP, GeoCLIP, CSP and PhilEO VHR as frozen feature extractors trained on other datasets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Predicting Internet Connectivity in Schools: A Feasibility Study Leveraging Multi-modal Data and Location Encoders in Low-Resource Settings." pith.science (2026). https://pith.science/paper/WFQ6KBXE

@misc{pith2026241212188,
  author       = {Pith},
  title        = {Pith review of: Predicting Internet Connectivity in Schools: A Feasibility Study Leveraging Multi-modal Data and Location Encoders in Low-Resource Settings},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WFQ6KBXE}},
  note         = {Machine review of arXiv:2412.12188}
}
read the original abstract

Internet connectivity in schools is critical to provide students with the digital literary skills necessary to compete in modern economies. In order for governments to effectively implement digital infrastructure development in schools, accurate internet connectivity information is required. However, traditional survey-based methods can exceed the financial and capacity limits of governments. Open-source Earth Observation (EO) datasets have unlocked our ability to observe and understand socio-economic conditions on Earth from space, and in combination with Machine Learning (ML), can provide the tools to circumvent costly ground-based survey methods to support infrastructure development. In this paper, we present our work on school internet connectivity prediction using EO and ML. We detail the creation of our multi-modal, freely-available satellite imagery and survey information dataset, leverage the latest geographically-aware location encoders, and introduce the first results of using the new European Space Agency phi-lab geographically-aware foundational model to predict internet connectivity in Botswana and Rwanda. We find that ML with EO and ground-based auxiliary data yields the best performance in both countries, for accuracy, F1 score, and False Positive rates, and highlight the challenges of internet connectivity prediction from space with a case study in Kigali, Rwanda. Our work showcases a practical approach to support data-driven digital infrastructure development in low-resource settings, leveraging freely available information, and provide cleaned and labelled datasets for future studies to the community through a unique collaboration between UNICEF and the European Space Agency phi-lab.

Figures

Figures reproduced from arXiv: 2412.12188 by the authors.

Figure 1
Figure 1. From left to right, Top: 3 connected schools in Rwanda, 2 connected schools in Botswana. Bottom: 3 [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Methodological workflow for connectivity prediction using engineered features and location encoder [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. PhilEO VHR sampling distribution [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Feature Importance. Left: Botswana Right: Rwanda [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: South-West Kigali. Left: Ground-Truth labels, Right: Model Predictions. predictions using the engineered feature space and highlight three locations in the South-Western part of the city (latitude, longitude bounds of (30.02,-2.01), (30.11,-1.93) respectively) of notew…
Figure 6
Figure 6. Figure 6: Depiction of 300m, 500m, 750m, 1000m, and 5000m buffer extents surrounding connected (green) and [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Average Test Run Accuracy and F1-Score for varying ML classifier with varying buffer size in Rwanda [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Botswana (left) and Rwanda (right) spatial distribution of connected (green triangle) and unconnected (red [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 34 canonical work pages

  1. [1]

    Arderne, C

    C. Arderne, C. Zorn, and C. et al. Nicolas. Predictive mapping of the global power system using open data. Nature Scientific Data, 7 0 (19), 2020. doi:https://doi.org/10.1038/s41597-019-0347-4

    work page doi:10.1038/s41597-019-0347-4 2020

  2. [2]

    atledge, G

    N. atledge, G. Cadamuro, and B. et al. de la Cuesta. Using machine learning to assess the livelihood impact of electricity access. Nature, 611: 0 491--495, 2022. doi:10.1038/s41586-022-05322-8

    doi:10.1038/s41586-022-05322-8 2022

  3. [3]

    Beck, Niklaus E

    Hylke E. Beck, Niklaus E. Zimmermann, Tim R. McVicar, Noemi Vergopolan, Alexis Berg, and Eric F. Wood. Present and future Köppen-Geiger climate classification maps at 1-km resolution . 7 2018. doi:10.6084/m9.figshare.6396959. URL https://figshare.com/articles/dataset/Present_and_future_K_ppen-Geiger_climate_classification_maps_at_1-km_resolution/6396959

    work page arXiv 2018

  4. [4]

    Bridging the Digital Divide, pages 283--298

    Anna Bon, Francis Saa-Dittoh, and Hans Akkermans. Bridging the Digital Divide, pages 283--298. Springer Nature Switzerland, Cham, 2024. ISBN 978-3-031-45304-5. doi:10.1007/978-3-031-45304-5_19. URL https://doi.org/10.1007/978-3-031-45304-5_19

    work page doi:10.1007/978-3-031-45304-5_19 2024

  5. [5]

    Gridded sex-disaggregated school-age population datasets for countries and dependent territories in africa in 2020, February 2022

    Maksym Bondarenko, Alessandro Sorichetta, Germ \'a n Vargas Mesa, Am \'e lie Gagnon A, and Andrew Tatem. Gridded sex-disaggregated school-age population datasets for countries and dependent territories in africa in 2020, February 2022. URL https://eprints.soton.ac.uk/454468/

    work page 2020

  6. [6]

    Adoption and Impact of Earth Observation for the 2030 Agenda for Sustainable Development

    Caribou Space . Adoption and Impact of Earth Observation for the 2030 Agenda for Sustainable Development. 2020

    work page 2020

  7. [7]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. A simple framework for contrastive learning of visual representations. In Hal Daumé III and Aarti Singh, editors, Proceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Research, pages 1597--1607. PMLR, 13--18 Jul 2020

    work page 2020

  8. [8]

    Functional Map of the World

    Gordon Christie , Neil Fendley , James Wilson , and Ryan Mukherjee . Functional Map of the World . arXiv e-prints, art. arXiv:1711.07846, November 2017. doi:10.48550/arXiv.1711.07846

    arXiv doi:10.48550/arxiv.1711.07846 2017

Show all 55 references
  1. [9]

    Gridded population of the world, version 4 (gpwv4): Population density

    CIESIN. Gridded population of the world, version 4 (gpwv4): Population density. New York: NASA Socioeconomic Data and Applications Center (SEDAC), 2018. doi:https://doi.org/10.7927/H49C6VHW

    2018 doi

  2. [10]

    Esa's ai4eo initiative: Bridging the gap between the ai and earth observation communities

    Annekatrien Debien, Mauro Casaburi, Grega Milcinski, and Marcello Maranesi. Esa's ai4eo initiative: Bridging the gap between the ai and earth observation communities. In 2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS, pages 251--253, 2021. doi:10.1109/I...

    2021

  3. [11]

    generalizable and accessible approach to machine learning with global satellite imagery

    Rolf E, Proctor J, Carleton T, Bolliger I, Shankar V, Ishihara, Recht B, and Hsiang S. generalizable and accessible approach to machine learning with global satellite imagery. Nature Communications, 20 0 (12), 2021. doi:10.1038/s41467-021-24638-z

    2021 doi

  4. [12]

    Elvidge, Kimgerly Baugh, Mikhail Zhizhin, Feng ChiHsu, and Tolottama Ghosh

    Christopher D. Elvidge, Kimgerly Baugh, Mikhail Zhizhin, Feng ChiHsu, and Tolottama Ghosh. Viirs night-time lights. International Journal of Remote Sensing, 38 0 (21): 0 5860--5879, 2017. doi:https://doi.org/10.1080/01431161.2017.1342050

    2017

  5. [13]

    Newcomers earth observation guide, 2020

    ESA. Newcomers earth observation guide, 2020. URL https://business.esa.int/newcomers-earth-observation-guide#: :text=EO

    2020

  6. [14]

    Object-based random forest classification for informal settlements identification in the middle east: Jeddah a case study

    Ahmad Fallatah, Simon Jones, and David Mitchell. Object-based random forest classification for informal settlements identification in the middle east: Jeddah a case study. International Journal of Remote Sensing, 41 0 (11): 0 4421--4445, 2020

    2020

  7. [15]

    Phileo bench: Evaluating geo-spatial foundation models, 2024

    Casper Fibaek, Luke Camilleri, Andreas Luyts, Nikolaos Dionelis, and Bertrand Le Saux. Phileo bench: Evaluating geo-spatial foundation models, 2024

    2024

  8. [16]

    Uncertainty-aware interpretable deep learning for slum mapping and monitoring

    Thomas Fisher, Harry Gibson, Yunzhe Liu, Moloud Abdar, Marius Posa, Gholamreza Salimi-Khorshidi, Abdelaali Hassaine, Yutong Cai, Kazem Rahimi, and Mohammad Mamouei. Uncertainty-aware interpretable deep learning for slum mapping and monitoring. Remote Sensing, 14 0 (13), 2022. ...

    2022 doi

  9. [17]

    Why do tree-based models still outperform deep learning on tabular data?, 2022

    Léo Grinsztajn, Edouard Oyallon, and Gaël Varoquaux. Why do tree-based models still outperform deep learning on tabular data?, 2022. URL https://arxiv.org/abs/2207.08815

    2022 arXiv

  10. [18]

    Improving remote sensing research and education in developing countries: Approaches and recommendations

    Barry Haack and Robert Ryerson. Improving remote sensing research and education in developing countries: Approaches and recommendations. International Journal of Applied Earth Observation and Geoinformation, 45: 0 77--83, 2016

    2016

  11. [19]

    Internet world stats - internet penetration in africa

    Internet World Stats . Internet world stats - internet penetration in africa. https://www.internetworldstats.com/stats1.htm, 2023. [Accessed 20-02-2024]

    2023

  12. [20]

    Measuring digital development: Facts and figures 2021

    ITU . Measuring digital development: Facts and figures 2021. www.itu.int/en/ITU-D/Statistics/Pages/facts/default.aspx, 2021. [Accessed 20-02-2024]

    2021

  13. [21]

    Achieving universal and meaningful digital connectivity setting a baseline and targets for 2030

    ITU and United Nations . Achieving universal and meaningful digital connectivity setting a baseline and targets for 2030. https://www.itu.int/itu-d/meetings/statistics/wp-content/uploads/sites/8/2022/04/UniversalMeaningfulDigitalConnectivityTargets2030_BackgroundPaper.pdf, 202...

    2022

  14. [22]

    Matthew Davis, David B

    Neal Jean, Marshall Burke, Michael Xie, W. Matthew Davis, David B. Lobell, and Stefano Ermon. Combining satellite imagery and machine learning to predict poverty. Science, 353 0 (6301): 0 790--794, 2016. doi:10.1126/science.aaf7894. URL https://www.science.org/doi/abs/10.1126/...

    2016 doi

  15. [23]

    Junteng Jia and Austion R. Benson. Residual correlation in graph neural network regression. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD '20, page 588–598, New York, NY, USA, 2020. Association for Computing Machinery....

    2020

  16. [24]

    Kennedy, J.R

    C.M. Kennedy, J.R. Oakleaf, D.M. Theobald, S. Baurch-Murdo, and J. Kiesecker. Managing the middle: A shift in conservation priorities based on the global human modification gradient. Global Change Biology, 00:1: 0 1--16, 2019. doi:doi:10.1111/gcb.14549

    2019 doi

  17. [25]

    SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery

    Konstantin Klemmer , Esther Rolf , Caleb Robinson , Lester Mackey , and Marc Ru wurm . SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery . arXiv e-prints, art. arXiv:2311.17179, November 2023. doi:10.48550/arXiv.2311.17179

    2023 doi arXiv

  18. [26]

    Martha Larson, Mohammad Soleymani, Guillaume Gravier, Bogdan Ionescu, and Gareth J.F. Jones. The benchmarking initiative for multimedia evaluation: Mediaeval 2016. IEEE MultiMedia, 24 0 (1): 0 93--96, 2017. doi:10.1109/MMUL.2017.9

    2016 doi

  19. [27]

    Pesaresi M. and P. Panagiotis. GHS-BUILT-C R2023A - GHS Settlement Characteristics, derived from Sentinel2 composite (2018) and other GHS R2023A data . European Commission, Joint Research Centre (JRC), PID: http://data.europa.eu/89h/3c60ddf6-0586-4190-854b-f6aa0edc2a30, 2023. ...

    2018 doi

  20. [28]

    CSP: Self-Supervised Contrastive Spatial Pre-Training for Geospatial-Visual Representations

    Gengchen Mai , Ni Lao , Yutong He , Jiaming Song , and Stefano Ermon . CSP: Self-Supervised Contrastive Spatial Pre-Training for Geospatial-Visual Representations . arXiv e-prints, art. arXiv:2305.01118, May 2023. doi:10.48550/arXiv.2305.01118

    2023 doi arXiv

  21. [29]

    GHSL Data Package 2023

    Schiavina Marcello, Melchiorri Michele, Pesaresi Martino, Politis Panagiotis, Carneiro Freire Sergio Manuel, Maffenini Luca, Florio Pietro, Ehrlich Daniele, Goch Katarzyna, Carioli Alessandra, Uhl Johannes, Tommasi Pierpaolo, and Kemper Thomas. GHSL Data Package 2023 . 0 (KJ-0...

    2023 doi

  22. [30]

    Establishing an operational and continuous monitoring of global built-up surfaces with the copernicus global human settlement layer

    Michele Melchiorri and Thomas Kemper. Establishing an operational and continuous monitoring of global built-up surfaces with the copernicus global human settlement layer. In 2023 Joint Urban Remote Sensing Event (JURSE), pages 1--4, 2023. doi:10.1109/JURSE57346.2023.10144201

    2023

  23. [31]

    Noor, V.A

    A.M. Noor, V.A. Alegana, and P.W. et al Gething. Using remotely sensed night-time light as a proxy for poverty in africa. Population Health Metrics, 6: 0 5, 2008. doi:10.1186/1478-7954-6-5

    2008 doi

  24. [32]

    Global fixed broadband and mobile network maps

    Ookla LLC. Global fixed broadband and mobile network maps. https://www.ookla.com/ookla-for-good/open-data, 2023. [Accessed 16-01-2024]

    2023

  25. [33]

    Sparse spatial autoregressions

    Kelley Pace and Ronald Barry. Sparse spatial autoregressions. Statistics and Probability Letters, 33 0 (3): 0 291--297, 1997. URL https://EconPapers.repec.org/RePEc:eee:stapro:v:33:y:1997:i:3:p:291-297

    1997

  26. [34]

    Pettersson, Mohammad Kakooei, Julia Ortheden, Fredrik D

    Markus B. Pettersson, Mohammad Kakooei, Julia Ortheden, Fredrik D. Johansson, and Adel Daoud. Time series of satellite imagery improve deep learning estimates of neighborhood-level poverty in africa. In Proceedings of the Thirty-Second International Joint Conference on Artific...

    2023 doi

  27. [35]

    Multi-source satellite imagery and point of interest data for poverty mapping in east java, indonesia: Machine learning and deep learning approaches

    Salwa Rizqina Putri, Arie Wahyu Wijayanto, and Setia Pramana. Multi-source satellite imagery and point of interest data for poverty mapping in east java, indonesia: Machine learning and deep learning approaches. Remote Sensing Applications: Society and Environment, 29: 0 10088...

    2023

  28. [36]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision. CoRR, abs/2103.00020...

    2021 arXiv

  29. [37]

    Learning transferable visual models from natural language supervision, 2021 b

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision, 2021 b

    2021

  30. [38]

    Artificial intelligence for social good: A survey

    Zheyuan Ryan Shi, Claire Wang, and Fei Fang. Artificial intelligence for social good: A survey. arXiv preprint arXiv:2001.01818, 2020

    2001 arXiv

  31. [39]

    Mcd12q1 modis/terra+aqua land cover type yearly l3 global 500m sin grid v006

    Damien Sulla-Menashe and Mark A Friedl. Mcd12q1 modis/terra+aqua land cover type yearly l3 global 500m sin grid v006. NASA EOSDIS Land Processes DAAC, 2023. doi:https://doi.org/10.5067/MODIS/MCD12Q1.006

    2023 doi

  32. [40]

    Detecting the linkage between arable land use and poverty using machine learning methods at global perspective

    Fuyou Tian, Bingfang Wu, Hongwei Zeng, Gary R Watmough, Miao Zhang, and Yurui Li. Detecting the linkage between arable land use and poverty using machine learning methods at global perspective. Geography and Sustainability, 3 0 (1): 0 7--20, 2022. ISSN 2666-6839. doi:https://d...

    2022 doi

  33. [41]

    Tingzon, A

    I. Tingzon, A. Orden, K. T. Go, S. Sy, V. Sekara, I. Weber, M. Fatehkia, M. Garc\' a-Herranz, and D. Kim. Mapping poverty in the philippines using machine learning, satellite imagery, and crowd-sourced geospatial information. The International Archives of the Photogrammetry, R...

    2019 doi

  34. [42]

    UNICEF . Giga. https://giga.global/ , year = 2023 , note = [Accessed 20-01-2024] ,, 2023 a

    2023

  35. [43]

    Project connect

    UNICEF . Project connect. https://projectconnect.unicef.org/map/countries, 2023 b . [Accessed 30-11-2024]

    2023

  36. [44]

    T wo thirds of the world’s school-age children have no internet access at home, new U N I C E F - I T U report says --- unicef.org

    UNICEF Europe and Central Asia . T wo thirds of the world’s school-age children have no internet access at home, new U N I C E F - I T U report says --- unicef.org. https://www.unicef.org/press-releases/two-thirds-worlds-school-age-children-have-no-internet-access-home-new-uni...

    2020

  37. [45]

    Van Dijk

    Jan A.G.M. Van Dijk. Jan van Dijk (2019/2020) The Digital Divide. Polity Press. 10 2019

    2019

  38. [46]

    The iNaturalist Species Classification and Detection Dataset

    Grant Van Horn , Oisin Mac Aodha , Yang Song , Yin Cui , Chen Sun , Alex Shepard , Hartwig Adam , Pietro Perona , and Serge Belongie . The iNaturalist Species Classification and Detection Dataset . arXiv e-prints, art. arXiv:1707.06642, July 2017. doi:10.48550/arXiv.1707.06642

    2017 doi arXiv

  39. [47]

    GeoCLIP: Clip-Inspired Alignment between Locations and Images for Effective Worldwide Geo-localization

    Vicente Vivanco Cepeda , Gaurav Kumar Nayak , and Mubarak Shah . GeoCLIP: Clip-Inspired Alignment between Locations and Images for Effective Worldwide Geo-localization . arXiv e-prints, art. arXiv:2309.16020, September 2023. doi:10.48550/arXiv.2309.16020

    2023 doi arXiv

  40. [48]

    The role of spatial heterogeneity in detecting urban slums

    Jiong Wang, Monika Kuffer, and Karin Pfeffer. The role of spatial heterogeneity in detecting urban slums. Computers, environment and urban systems, 73: 0 95--107, 2019

    2019

  41. [49]

    Deep cosine metric learning for person re-identification

    Nicolai Wojke and Alex Bewley. Deep cosine metric learning for person re-identification. In 2018 IEEE Winter Conference on Applications of Computer Vision (WACV), pages 748--756, 2018. doi:10.1109/WACV.2018.00087

    2018

  42. [50]

    Semantic segmentation of slums in satellite images using transfer learning on fully convolutional neural networks

    Michael Wurm, Thomas Stark, Xiao Xiang Zhu, Matthias Weigand, and Hannes Taubenb \"o ck. Semantic segmentation of slums in satellite images using transfer learning on fully convolutional neural networks. ISPRS journal of photogrammetry and remote sensing, 150: 0 59--69, 2019

    2019

  43. [51]

    Combining night time lights in prediction of poverty incidence at the county level

    Jianbin Xu, Jie Song, Baochao Li, Dan Liu, and Xiaoshu Cao. Combining night time lights in prediction of poverty incidence at the county level. Applied Geography, 135: 0 102552, 2021. ISSN 0143-6228. doi:https://doi.org/10.1016/j.apgeog.2021.102552. URL https://www.sciencedire...

    2021

  44. [52]

    C. Yeh, A. Perez, and A. et al. Driscoll. Using publicly available satellite imagery and deep learning to understand economic well-being in africa. Nature Communications, 11: 0 2583, 2020. doi:10.1038/s41467-020-16185-w. URL https://www.nature.com/articles/s41467-020-16185-w#citeas

    2020 doi

  45. [53]

    Gps2vec: Towards generating worldwide gps embeddings

    Yifang Yin, Zhenguang Liu, Ying Zhang, Sheng Wang, Rajiv Ratn Shah, and Roger Zimmermann. Gps2vec: Towards generating worldwide gps embeddings. In Proceedings of the 27th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems, SIGSPATIAL '19, pag...

    2019

  46. [54]

    L. Yu. Meta-discoveries from a synthesis of satellite-based land-cover mapping research. International Journal of Remote Sensing, 35 0 (4573), 2014

    2014

  47. [55]

    Esa worldcover 10 m 2021 v200, October 2022

    Daniele Zanaga, Ruben Van De Kerchove, Dirk Daems, Wanda De Keersmaecker, Carsten Brockmann, Grit Kirches, Jan Wevers, Oliver Cartus, Maurizio Santoro, Steffen Fritz, Myroslava Lesiv, Martin Herold, Nandin-Erdene Tsendbazar, Panpan Xu, Fabrizio Ramoino, and Olivier Arino. Esa ...

    2021 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.