Learning to Holistically Detect Bridges From Large-Size VHR Remote Sensing Imagery
Open accessThe DOI matches its Crossref record; 68 of the 68 references it cites were found in a registry or an index; 6 of the 64 sources its claims cite were checked for support.
Abstract
Bridge detection in remote sensing images (RSIs) plays a crucial role in various applications, but it poses unique challenges compared to the detection of other objects. In RSIs, bridges exhibit considerable variations in terms of their spatial scales and aspect ratios. Therefore, to ensure the visibility and integrity of bridges, it is essential to perform holistic bridge detection in large-size very-high-resolution (VHR) RSIs. However, the lack of datasets with large-size VHR RSIs limits the deep learning algorithms' performance on bridge detection. Due to the limitation of GPU memory in tackling large-size images, deep learning-based object detection methods commonly adopt the cropping strategy, which inevitably results in label fragmentation and discontinuous prediction. To ameliorate the scarcity of datasets, this paper proposes a large-scale dataset named GLH-Bridge comprising 6,000 VHR RSIs sampled from diverse geographic locations across the globe. These images encompass a wide range of sizes, varying from 2,048 × 2,048 to 16,384 × 16,384 pixels, and collectively feature 59,737 bridges. These bridges span diverse backgrounds, and each of them has been manually annotated, using both an oriented bounding box (OBB) and a horizontal bounding box (HBB). Furthermore, we present an efficient network for holistic bridge detection (HBD-Net) in large-size RSIs. The HBD-Net presents a separate detector-based feature fusion (SDFF) architecture and is optimized via a shape-sensitive sample re-weighting (SSRW) strategy. The SDFF architecture performs inter-layer feature fusion (IFF) to incorporate multi-scale context in the dynamic image pyramid (DIP) of the large-size image, and the SSRW strategy is employed to ensure an equitable balance in the regression weight of bridges with various aspect ratios. Based on the proposed GLH-Bridge dataset, we establish a bridge detection benchmark including the OBB and HBB tasks, and validate the effectiveness of the proposed HBD-Net. Additionally, cross-dataset generalization experiments on two publicly available datasets illustrate the strong generalization capability of the GLH-Bridge dataset.
Overview
CoChat Summary
TL;DR
The paper introduces GLH-Bridge, a 6,000-image large-size VHR remote sensing dataset with 59,737 bridge annotations, and HBD-Net, a network for holistic bridge detection using separate detector-based feature fusion and shape-sensitive sample re-weighting. On the GLH-Bridge benchmark, HBD-Net achieves 35.35% mAP for OBB and 34.49% mAP for HBB, with strong AP on large/huge bridges and reported cross-dataset generalization.
Key findings
- GLH-Bridge contains 6,000 VHR RSIs from over 400 cities/regions, with image sizes from 2,048 × 2,048 to 16,384 × 16,384 pixels, GSD 0.3–1.0 m, and 59,737 bridges manually annotated with both OBB and HBB; the split is 3,613 training, 1,194 validation, and 1,193 testing images.
- HBD-Net achieves the best OBB result on GLH-Bridge with 35.35% mAP, 28.69% AP75, 33.47% AP_lg, and 20.61% AP_hg; for HBB it achieves 34.49% mAP, 35.21% AP_lg, and 35.59% AP_hg.
- Ablations show SDFF without cross-scale-transfer distillation and IFF improves AP_hg by 8.57% over baseline; adding cross-scale-transfer distillation adds 3.4% AP_hg; IFF reaches 28.06% AP75 and 17.30% AP_hg. SSRW on the bottom-layer detector improves AP_lg by 1.37% and AP_hg by 3.30%, and the final HBD-Net improves AP75 by 2.40%, AP_lg by 4.17%, and AP_hg by 14.93% over baseline.
- Cross-dataset generalization experiments report improvements of 3.79% on DOTA-v1.0 and 0.86% on DIOR-R; on DOTA-v1.0, HBD-Net reaches 46.11% mAP and 56.02% AP_BR.
Methods
- Built GLH-Bridge from Google Earth and MapBox imagery collected 2019–2022, using NBI metadata and global city/river sampling; bridges were manually annotated with OBB and HBB under a three-stage annotation process with expert review.
- Proposed HBD-Net with a separate detector-based feature fusion (SDFF) architecture over a dynamic image pyramid (DIP), inter-layer feature fusion (IFF), cross-scale-transfer distillation, and a shape-sensitive sample re-weighting (SSRW) strategy for aspect-ratio-sensitive regression weighting.
- Established OBB and HBB benchmarks on GLH-Bridge against 18 object detection methods using MMRotate/MMDetection, with AP metrics including length-based AP_sh, AP_md, AP_lg, and AP_hg; experiments used one Tesla V100 GPU with 16 GB memory, batch size 1 for HBD-Net, downsampling ratio 2.0, and 1024×1024 cropping windows with 200-pixel overlap for comparison methods.
Limitations
- The paper does not include a formal limitations section; it states future work will enrich GLH-Bridge in sample volume and sub-category annotation, and generalize HBD-Net to multi-class object detection in large-size images.
- Evident from the text: GLH-Bridge is single-class (bridge only) and lacks sub-category labels, and the SSRW strategy is not used for the HBB task because horizontal-box aspect ratio does not accurately reflect whether bridges are elongated.
- Evident from Table 6: zero-shot transfer from GLH-Bridge to DOTA-v1.0 and DIOR-R yields 15.78% and 15.01% AP, respectively, lower than in-domain training on those datasets (45.76% and 19.88%), indicating a generalization gap despite the paper’s reported relative improvements.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
1 Valid · 5 Not valid · 41 Not checked · 128 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“although numerous popular datasets for object detection in RSIs have been created [6], [7], [8], [17], the quantity of bridges within these datasets is limited.”
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 1 INTRODUCTION
What the sources say
- Partly supportedcites [6] X. Sun, P. Wang, Z. Yan, F. Xu, R. Wang, W. Diao, J. Chen, J. Li, Y. Feng, T. Xu et al., "Fair1m: A benchmark dataset for fine-grained object recognition in high-resolution remote sensing imagery," ISPRS Journal of Photogrammetry and Remote Sensing, vol. 184, pp. 116-130, 2022. sourceExperimental — accuracy not yet measured
FAIR1M (Sec. 3.4, 'Characteristics of the Dataset') does state that many existing remote-sensing object detection datasets have 'the number of instances and categories ... relatively small', which supports limited instance quantities only at a general level. Its sole mention of bridges (Sec. 3.2, 'Category Design') is that 'most existing datasets pay more attention to static objects, such as bridges, baseball fields, storage tanks...' and that these datasets 'lack fine-grained information' — a comment on annotation granularity, not on how few bridge instances they contain. No passage in our copy (searched: bridge instances/quantity, Table 1 dataset comparison, category distributions) reports that the number of bridges in popular RSI detection datasets is limited, so that specific part of the claim is unsupported by this source. (LLM, conf=0.65)
- Could not verifycites [7] G.-S. Xia, X. Bai, J. Ding, Z. Zhu, S. Belongie, J. Luo, M. Datcu, M. Pelillo, and L. Zhang, "Dota: A large-scale dataset for object detection in aerial images," in IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 3974-3983. sourceExperimental — accuracy not yet measured
Our copy of DOI 10.1109/cvpr.2018.00418 contains only the IRIS/ARCA institutional repository landing-page records (title, authors, conference label, handle https://hdl.handle.net/10278/3717302, file listing 'CVPR 2018.pdf, 1.24 MB', ASN simulation boilerplate) — no body text of the DOTA paper itself. Searches for the dataset's object categories and bridge annotations, and for statements about existing remote-sensing detection datasets, returned only these same metadata pages (pages 1-6 of the indexed result, sections null and '2018 IEEE Conference on CVPR (CVPR 2018)'). Nothing in our copy states or contradicts that the number of bridges in popular RSI object-detection datasets is limited, so the claim cannot be checked against this source. (LLM, conf=0.75)
- Partly supportedcites [8] J. Ding, N. Xue, G.-S. Xia, X. Bai, W. Yang, M. Yang, S. Belongie, J. Luo, M. Datcu, M. Pelillo, and L. Zhang, "Object detection in aerial images: A large-scale benchmark and challenges," IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 1-1, 2021. sourceExperimental — accuracy not yet measured
Our copy supports the general half of the sentence: Sec. 2.2 (p. 3) and the Introduction (p. 2) survey numerous public aerial/RSI object-detection datasets (Table 2, pp. 3-4, lists ~20) and state that they 'contain a limited number of instances' / 'have only limited number of samples'. But no passage attributes limited bridge quantities to those datasets specifically — Sec. 2.2 characterises prior datasets as building, vehicle, ship and plane datasets, and Table 2 gives only total instance counts, never per-category bridge counts. The only bridge numbers in our copy are DOTA's own (Sec. 4.7.1, p. 9: 3,760 / 3,804 / 21,433), which is the authors' dataset, not the popular datasets the claim refers to. (LLM, conf=0.78)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
None of the sources cited for this claim have been checked yet.
In the paper
“Despite these datasets containing the bridge category, they have limited quantities of bridge instances.”
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 2.1.1 Multi-Class Datasets for Bridge Detection
What the sources say
- Not checked yetcites [5] G. Cheng, J. Han, P. Zhou, and L. Guo, "Multi-class geospatial object detection and geographic image classification based on collection of part detectors," ISPRS Journal of Photogrammetry and Remote Sensing, vol. 98, pp. 119-132, 2014. source
Checked before — its earlier verdict is held back until a new check runs.
- Could not verifycites [7] G.-S. Xia, X. Bai, J. Ding, Z. Zhu, S. Belongie, J. Luo, M. Datcu, M. Pelillo, and L. Zhang, "Dota: A large-scale dataset for object detection in aerial images," in IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 3974-3983. sourceExperimental — accuracy not yet measured
Our copy of the DOTA CVPR 2018 paper holds only the IRIS/ARCA repository catalogue record — title/author list, venue and DOI block, file metadata ('CVPR 2018.pdf', 1.24 MB, Adobe PDF) and Italian ASN-simulation boilerplate — with no body text (sections, pages absent or null). Four searches ('bridge category limited number of instances', 'object categories bridge small rare classes distribution', '2806 images 18828 instances 15 categories', 'existing aerial image datasets comparison small objects oriented bounding boxes') returned nothing but that metadata, so no passage states or contradicts that existing multi-class datasets containing the bridge category have limited bridge instance counts. The decisive comparison of prior datasets' category inventances would sit in the paper's related-work/dataset-description sections, which our copy does not contain. (LLM, conf=0.70)
- Could not verifycites [17] C. Liu, J. Yang, J. Ou, and D. Fan, "Offshore bridge detection in polarimetric sar images based on water network construction using markov tree," Remote Sensing, vol. 14, no. 16, p. 3888, 2022. sourceExperimental — accuracy not yet measured
Our copy of Liu et al. (2022, Remote Sensing) is a method paper: the Introduction (Section 1) reviews existing bridge detection approaches (water-land segmentation plus geometric/spatial or intensity features) and states the difficulties of SAR speckle, coastal strong scatterers and terrain morphology, but nowhere does it survey or characterise existing multi-class datasets or the number of bridge instances they contain. The only dataset statements concern the authors' own three test scenes (Section 4, Table 1/Section 4.4: 16, 5 and 11 bridges, ground truth drawn with reference to Google Earth) and a Data availability statement directing readers to contact the author. Searched for 'limited quantities of bridge instances', 'multi-class dataset bridge category', 'no public SAR dataset / we build our own labeled dataset', and 'deep learning object detection dataset bridge samples' — no passage addresses the claim's point, so it can be neither supported nor contradicted. (LLM, conf=0.70)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not checked yet
None of the sources cited for this claim have been checked yet.
In the paper
“Given the constraints of GPU memory, mainstream deep learning-based object detection methods [14], [15], [18], [19], [20] commonly employ cropping strategies [7], [21].”
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 1 INTRODUCTION
What the sources say
- Not checked yetcites [14] X. Yang, G. Zhang, X. Yang, Y. Zhou, W. Wang, J. Tang, T. He, and J. Yan, "Detecting rotated objects as gaussian distributions and its 3-d generalization," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 4, pp. 4335-4354, 2022. source
Checked before — its earlier verdict is held back until a new check runs.
- Not checked yetcites [15] G. Nie and H. Huang, "Multi-oriented object detection in aerial images with double horizontal rectangles," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 4, pp. 4932- 4944, 2022. source
Checked before — its earlier verdict is held back until a new check runs.
- Could not verifycites [18] K. Li, G. Wan, G. Cheng, L. Meng, and J. Han, "Object detection in optical remote sensing images: A survey and a new benchmark," ISPRS Journal of Photogrammetry and Remote Sensing, vol. 159, pp. 296-307, 2020. sourceExperimental — accuracy not yet measured
Our copy of the survey describes mainstream deep learning-based detectors (Sec. 2.2.1–2.2.2, pp. 4–6; Sec. 5.1, pp. 12–13, where 12 'representative deep learning based object detection methods, which are widely used for object detection in natural scene images and earth observation images' are benchmarked) and the datasets used (Sec. 3.1, p. 8), but no passage links GPU memory limits to a cropping strategy. Searches for 'GPU memory', 'cropping/crop', 'patches/sub-images' and 'sliding window' returned only the sliding-window mention for OverFeat (p. 6) and super-pixel local patches in Yu et al. (2015) (Sec. 3.2, p. 9), neither of which states the claim's causal point. The paper is not truncated, so the asserted reason (GPU memory constraints) and practice (common cropping) are simply not addressed in our copy. (LLM, conf=0.68)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not checked yet
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Experimental results on GLH-Bridge demonstrate the outstanding performance of our proposed HBD-Net.”
- Also stated in: § Abstract · plain
“validate the effectiveness of the proposed HBD-Net”
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 1 INTRODUCTION
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Abstract · plain
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“The separate detector-based feature fusion (SDFF) architecture, when applied to the dynamic image pyramid (DIP), demonstrates an efficient approach for processing large-size images with minimal resource consumption.”
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 1 INTRODUCTION
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
85 more claims
Create a free account to see all 175 claims and to verify the 57 references still open.
- Probably not a substantial claim. 54
- It may state two separate claims as one. 40
- The quote only partly states this claim. 3
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 3.2.1 Annotation Criteria
“Objects deemed to be non-bridges or bridges presenting challenges in labeling are deliberately omitted from the labeling process, such as two terminal connections shown in Fig. 3(a).Fig. 3(b) shows a road across the water with excessive curvature or an irregular shape that will not be labeled.”
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 6 CONCLUSION
“Fig. 3. Examples of labeling according to the criteria.(a) Roads across water with excessive curvature or an irregular shape are not labeled.(b) Two terminal connections are not labeled.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 2.1.2 Specialized Datasets for Bridge Detection
“This dataset comprises 500 images with a consistent size of 4,800 × 2,843 pixels.”
§ Learning to Holistically Detect Bridges from Large-Size VHR Remote Sensing Imagery > 6 CONCLUSION
“| Bridges Dataset [11] | 208 | 4,800×2,843 | 0.5 | 322 | HBB | ✓ | Google Earth |”
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472
ImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333
Microsoft COCO: Common Objects in Context
Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., & Zitnick, C. L. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10602-1_48 · Cited by 45,460
Feature Pyramid Networks for Object Detection
Lin, T.-Y., Dollar, P., Girshick, R., He, K., Hariharan, B., & Belongie, S. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.106 · Cited by 30,939 · cited 1×
“…complementary cues to feature fusion, we devise an Inter-Layer Feature Fusion (IFF) module to enable bidirectional feature sharing within the SDFF. Similar to the basic feature extractors (e.g., Resnet [49] followed by FPN [50]), this paper recommends extracting feature pyramids from the DIP.”
MethodGirshick, R. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.169 · Cited by 28,105
The pascal visual object classes (VOC) challenge
Everingham, M., Van Gool, L., Williams, C. K. I., Winn, J., & Zisserman, A. (2010) · International Journal of Computer Vision · doi:10.1007/s11263-009-0275-4 · Cited by 19,599
Focal Loss for Dense Object Detection
Lin, T.-Y., Goyal, P., Girshick, R., He, K., & Dollar, P. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2858826 · Cited by 9,556
DOTA: A Large-Scale Dataset for Object Detection in Aerial Images
Xia, G.-S., Bai, X., Ding, J., Zhu, Z., Belongie, S., Luo, J., Datcu, M., Pelillo, M., & Zhang, L. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00418 · Cited by 3,414
Self-citationZhang, S., Chi, C., Yao, Y., Lei, Z., & Li, S. Z. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00978 · Cited by 2,354
Cited by
Decentralized Federated Learning: A Survey and Perspective
Yuan, L., Wang, Z., Sun, L., Yu, P. S., & Brinton, C. G. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3407584 · Cited by 290
Boundary-Aware Feature Fusion With Dual-Stream Attention for Remote Sensing Small Object Detection
Song, J., Zhou, M., Luo, J., Pu, H., Feng, Y., Wei, X., & Jia, W. (2024) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2024.3514376 · Cited by 196
Self-citationA survey of small object detection based on deep learning in aerial images
Hua, W., & Chen, Q. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-025-11150-9 · Cited by 166
Oriented object detection in optical remote sensing images using deep learning: a survey
Wang, K., Wang, Z., Li, Z., Su, A., Teng, X., Pan, E., Liu, M., & Yu, Q. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-025-11256-0 · Cited by 118
Li, Y., Wang, L., Wang, T., Yang, X., Luo, J., Wang, Q., Deng, Y., Wang, W., Sun, X., Li, H., et al. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3508072 · Cited by 110
Self-citation
Metadata
- Authors
- Li, Y., Luo, J., Zhang, Y., Tan, Y., Yu, J.-G., & Bai, S.
- Affiliations
- Wuhan University (CN)
- Huazhong University of Science and Technology (CN)
- South China University of Technology (CN)
- ByteDance (China) (CN)
- Venue
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Published
- 2024-12
- Volume
- 46
- Issue
- 12
- Pages
- 11507-11523
- ISSN
- 0162-8828 (ISSN-L)
- 2160-9292
- 1939-3539
- OpenAlex
- W4396241435
- Cited by
- 77
- References
- 66
- Citations by year
- 2026: 34 · 2025: 34 · 2024: 9
- FWCI
- 6.65
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Pattern Analysis and Machine Intelligence (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Natural Science Foundation of China (42371321, 42030102)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Automated Road and Building Extraction
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Remote-Sensing Image Classification
- Keywords
- bridge detection, oriented bounding boxes, feature fusion, aspect ratio, sample reweighting, object detection, cross-dataset generalization, infrastructure monitoring
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-11-19, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-19 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.