Image-Based Lunar Hazard Detection in Low Illumination Simulated Conditions via Vision Transformers
Open accessThe DOI matches its Crossref record; 32 of the 36 references it cites were found in a registry or an index; 2 of the 5 sources its claims cite were checked for support.
Abstract
Hazard detection is fundamental for a safe lunar landing. State-of-the-art autonomous lunar hazard detection relies on 2D image-based and 3D Lidar systems. The lunar south pole is challenging for vision-based methods. The low sun inclination and the terrain rich in topographic features create large areas in shadow, hiding the terrain features. The proposed method utilizes a vision transformer (ViT) model, which is a deep learning architecture based on the transformer blocks used in natural language processing, to solve this problem. Our goal is to train the ViT model to extract terrain features information from low-light RGB images. The results show good performances, especially at high altitudes, beating the UNet, one of the most popular convolutional neural networks, in every scenario.
Overview
CoChat Summary
TL;DR
The paper trains a vision transformer (ViT) for image-based lunar hazard detection under low-illumination simulated conditions and compares it with a UNet. The ViT outperforms the UNet on IoU and precision, especially at higher altitudes, though it is slower and evaluated only on synthetic Blender-rendered data.
Key findings
- The ViT model outperformed the UNet for both IoU and precision on the test set, while recall values were very similar; the authors conclude the UNet classification is biased toward hazards.
- Both models performed better at higher altitudes, with ViT performance best between 5000 and 9000 m; on a “cleaned” test set that discarded 4891 of 25,000 images with no information, both models improved mainly at lower altitudes, and the ViT IoU distribution peaked around 80% for 5000–9000 m.
- Computational comparison: ViT mean inference time was 0.346 s per image (max 0.500 s, min 0.318 s), versus UNet mean 0.012 s (max 0.054 s, min 0.006 s).
- The ViT had 105 million parameters versus 17 million for the UNet.
Methods
- The authors used a ViT-Base encoder pre-trained on ImageNet and trained the decoder from scratch for semantic segmentation of safe versus hazardous lunar terrain; training used ADAM with learning rate 1 × 10−5, cross-entropy loss, batch size 5, and 100 epochs.
- The dataset contains 125,000 Blender-rendered RGB images of the lunar south pole (90 × 90 km area) at 384 × 384 resolution, generated from a NASA digital terrain model with a sun lamp inclination of 1.5 deg and camera altitudes from 500 m to 10 km; labels mark pixels safe if slope is less than 8 deg and roughness less than 10%.
- The dataset was split 60% training, 20% validation, and 20% testing; the test set contained 25,000 images. Evaluation used precision, recall, and IoU, and the UNet was trained on the same dataset and augmentation with batch size 20 and 100 epochs.
Limitations
- The paper states that no noise effects were considered, the Blender camera model differs from a real CMOS sensor, and the authors lacked real measurements or accurate lab data to create a realistic photometric model; therefore they cannot claim how the dataset compares with real data.
- The paper also notes the model does not consider Moon curvature, cannot substitute an active sensor like LiDAR, and is not optimized for real-world deployment.
- Evident from the text: evaluation is entirely on simulated Blender-rendered images with no real lunar imagery or real-camera validation, and the ViT’s much slower inference time (0.346 s vs. 0.012 s per image) limits practical deployment compared with the UNet.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures

![Figure 2 The network architecture of ViT [27].](https://figures.cochat.ai/paper-figures/0e2f7101a3c744239b12941ebf16a4c4/image-based-lunar-hazard-detection-in-low-illumination-simulated-conditions-via-figure-2.jpg)
![Figure 3 Vision transformer network architecture. This picture is a modified version of the architecture overview presented in [26].](https://figures.cochat.ai/paper-figures/c24d72fcb6eb4d2092ab727a207bc1a5/image-based-lunar-hazard-detection-in-low-illumination-simulated-conditions-via-figure-3.jpg)






Verification
Claims the paper makes
2 Valid · 5 Not valid · 2 Not checked · 33 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“For image-based hazard detection in particular, some machine learning classifiers, such as perceptrons [15], have been explored. Still, these approaches generally require the manual input of relevant features of the surface.”
§ Image-Based Lunar Hazard Detection in Low Illumination Simulated Conditions via Vision Transformers > 1. Introduction
What the sources say
- Supportedcites [15] Lunghi, 2016, A multilayer perceptron hazard detector for vision-based autonomous planetary landing, Adv. Astronaut. Sci. source
Our copy is exactly the kind of classifier the citing sentence names: a multilayer-perceptron (feed-forward back-propagation ANN) hazard detector operating on a single VIS image (Abstract; §2.3). Its input is not raw imagery but a hand-assembled vector of surface-related indexes — µ, σ, Grad and LoG at three scales plus sun inclination angle (§2.2, 'Information Extraction and ANN Input Assembly'), selected by the authors from 24 candidates in a try-and-error process (§2.2). The Conclusion states the general point the claim attributes: 'the selection of the indexes remains dependent on the user choice', with automatic representation learning (autoencoders, CNN, SOM) named only as future work. The citing sentence's mild generalization from this one perceptron system to 'these approaches generally' is the only stretch, and the paper itself frames index selection as a persistent limitation rather than a one-off. (LLM, conf=0.85)
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.
- Supported
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Both models’ performance mainly benefits by filtering out the images with no information at lower altitudes.”
§ Image-Based Lunar Hazard Detection in Low Illumination Simulated Conditions via Vision Transformers > 5. Results
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
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
“However, the recall values for both models are very similar”
§ Image-Based Lunar Hazard Detection in Low Illumination Simulated Conditions via Vision Transformers > 5. Results
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.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Apparently, both models perform better at higher altitudes.”
§ Image-Based Lunar Hazard Detection in Low Illumination Simulated Conditions via Vision Transformers > 5. Results
hedgedstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It is important to note that images without information poison the metrics since the camera sensor did not gather enough light.”
§ Image-Based Lunar Hazard Detection in Low Illumination Simulated Conditions via Vision Transformers > 6. Conclusions
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.
16 more claims
Create a free account to see all 42 claims and to verify the 1 reference still open.
- It may state two separate claims as one. 8
- Probably not a substantial claim. 15
- The quote only partly states this claim. 1
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
§ Image-Based Lunar Hazard Detection in Low Illumination Simulated Conditions via Vision Transformers > 5. Results
“The test set is 20% of randomly selected pictures in the training set, for a total of 25,000 images spanning from a minimum height of 500 m to a maximum of 10 km.”
§ Image-Based Lunar Hazard Detection in Low Illumination Simulated Conditions via Vision Transformers > 2. Method > 2.2. Dataset
“Then, 60% of the dataset is randomly selected for the training, 20% is selected for the validation, and the last 20% is used for testing.”
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
U-Net: Convolutional Networks for Biomedical Image Segmentation
Ronneberger, O., Fischer, P., & Brox, T. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-24574-4_28 · Cited by 99,404 · cited 1×
“Based on a deep convolutional encoder/decoder (DCE) structure, this architecture is known for its versatility and high accuracy in various applications [20–22].”
MethodExploiting Generative AI to Scale up Intelligent Tutoring Systems
Jakubův, J., Chvalovský, K., Goertzel, Z., Kaliszyk, C., Olšák, M., Piotrowski, B., Schulz, S., Suda, M., & Urban, J. (2023) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.itp.2023.19 · Cited by 76,658
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
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 47,472
A Computational Approach to Edge Detection
Canny, J. (1986) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.1986.4767851 · Cited by 30,004 · cited 1×
“Among these methods, there are the Canny [9,10] and Sobel [11] algorithms, which detect the edges in an image, and the various corner detection algorithms [12,13], which can provide good repeatability along with localization [14].”
MethodA Combined Corner and Edge Detector
Harris, C., & Stephens, M. (1988) · Procedings of the Alvey Vision Conference 1988 · doi:10.5244/c.2.23 · Cited by 12,901 · cited 1×
“Among these methods, there are the Canny [9,10] and Sobel [11] algorithms, which detect the edges in an image, and the various corner detection algorithms [12,13], which can provide good repeatability along with localization [14].”
MethodTransformers in Vision: A Survey
Khan, S., Naseer, M., Hayat, M., Zamir, S. W., Khan, F. S., & Shah, M. (2022) · ACM Computing Surveys · doi:10.1145/3505244 · Cited by 3,595
H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation From CT Volumes
Li, X., Chen, H., Qi, X., Dou, Q., Fu, C.-W., & Heng, P.-A. (2018) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2018.2845918 · Cited by 2,473
Vision Transformers for Dense Prediction
Ranftl, R., Bochkovskiy, A., & Koltun, V. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.01196 · Cited by 2,373 · cited 3×
“A more detailed description of the decoder can be found in [26].”
BackgroundMethodInfluentialRobust image corner detection through curvature scale space
Mokhtarian, F., & Suomela, R. (1998) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/34.735812 · Cited by 548 · cited 1×
“Among these methods, there are the Canny [9,10] and Sobel [11] algorithms, which detect the edges in an image, and the various corner detection algorithms [12,13], which can provide good repeatability along with localization [14].”
MethodA Descriptive Algorithm for Sobel Image Edge Detection
Vincent, O., & Folorunso, O. (2009) · InSITE Conference · doi:10.28945/3351 · Cited by 534 · cited 1×
“Among these methods, there are the Canny [9,10] and Sobel [11] algorithms, which detect the edges in an image, and the various corner detection algorithms [12,13], which can provide good repeatability along with localization [14].”
MethodIllumination conditions of the lunar polar regions using LOLA topography
Mazarico, E., Neumann, G. A., Smith, D. E., Zuber, M. T., & Torrence, M. H. (2011) · Icarus · doi:10.1016/j.icarus.2010.10.030 · Cited by 376 · cited 1×
“The scene’s illumination is provided by a “sun lamp” with an inclination of 1.5 deg, with the terrain rotated randomly around the azimuth axis [32].”
InfluentialBarker, M. K., Mazarico, E., Neumann, G. A., Smith, D. E., Zuber, M. T., & Head, J. W. (2021) · Planetary and Space Science · doi:10.1016/j.pss.2020.105119 · Cited by 133 · cited 1×
“Within Blender, we load a digital terrain model (DTM) of the area of interest provided by NASA (https://pgda.gsfc.nasa.gov/products/81, 10 October 2022) [31].”
MethodAutonomous Precision Landing and Hazard Detection and Avoidance Technology (ALHAT)
Epp, C. D., & Smith, T. B. (2007) · Proceedings - IEEE Aerospace Conference · doi:10.1109/aero.2007.352724 · Cited by 110 · cited 1×
“NASA’s ALHAT (Autonomous Precision Landing and Hazard Detection and Avoidance Technology) [1–3] program is developing new technologies for lunar landing guidance, navigation, and control to support both manned and unmanned missions.”
BackgroundGuidance Summary and Assessment of the Chang’e-3 Powered Descent and Landing
Li, S., Jiang, X., & Tao, T. (2016) · Journal of Spacecraft and Rockets · doi:10.2514/1.a33208 · Cited by 86 · cited 1×
“D) laser-elevation-image-based [5,6].”
Deep Learning Crater Detection for Lunar Terrain Relative Navigation
Downes, L., Steiner, T. J., & How, J. P. (2020) · AIAA Scitech 2020 Forum · doi:10.2514/6.2020-1838 · Cited by 57
Guidance Navigation and Control for Chang’E-5 Powered Descent
Zhang, H., Li, J., Wang, Z., & YiFeng, G. U. A. N. (2021) · Space Science & Technology · doi:10.34133/2021/9823609 · Cited by 56 · cited 1×
“The following Chang’e 4 and 5 landers used a similar design to Chang’e 3 [7,8], confirming the robustness of the design.”
MethodInfluentialDisk-resolved photometric modeling and properties of asteroid (101955) Bennu
Golish, D. R., DellaGiustina, D. N., Li, J.-Y., Clark, B. E., Zou, X.-D., Smith, P. H., Rizos, J. L., Hasselmann, P. H., Bennett, C. A., Fornasier, S., et al. (2021) · Icarus · doi:10.1016/j.icarus.2020.113724 · Cited by 47 · cited 1×
“However, this method requires real measurements or very accurate lab data to create a realistic photometric model [35], which the authors do not have access to at this time.”
MethodThe encoding method of position embeddings in vision transformer
Jiang, K., Peng, P., Lian, Y., & Xu, W. (2022) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2022.103664 · Cited by 39
Real-Time Hazard Detection and Avoidance Demonstration for a Planetary Lander
Epp, C. D., Robertson, E., & Carson, J. M. (2014) · AIAA SPACE Conference & Exposition · doi:10.2514/6.2014-4312 · Cited by 30 · cited 1×
“NASA’s ALHAT (Autonomous Precision Landing and Hazard Detection and Avoidance Technology) [1–3] program is developing new technologies for lunar landing guidance, navigation, and control to support both manned and unmanned missions.”
BackgroundA Deep Learning Approach to Hazard Detection for Autonomous Lunar Landing
Moghe, R., & Zanetti, R. (2020) · The Journal of the Astronautical Sciences · doi:10.1007/s40295-020-00239-8 · Cited by 27 · cited 1×
“In the past, the authors and other research groups explored the capabilities of the UNet [16] architecture applied to the lunar landing hazard detection problem [17–19].”
BackgroundA multilayer perceptron hazard detector for vision-based autonomous planetary landing
Lunghi, P., Ciarambino, M., & Lavagna, M. (2016) · Advances in Space Research · doi:10.1016/j.asr.2016.04.012 · Cited by 26 · cited 1×
“For image-based hazard detection in particular, some machine learning classifiers, such as perceptrons [15], have been explored.”
BackgroundRCDPT: Radar-Camera Fusion Dense Prediction Transformer
Lo, C.-C., & Vandewalle, P. (2023) · ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp49357.2023.10096129 · Cited by 18 · cited 1×
“The readout token is generally responsible for aggregating information from other tokens [30].”
BackgroundVision based Hazard detection and obstacle Avoidance for planetary landing
Mahmood, W., & Shah, S. (2009) · Joint International Workshop on Nonlinear Dynamics and Synchronization and International Symposium on Theoretical Electrical Engineering (INDS & ISTET) · doi:10.1109/inds.2009.5227995 · Cited by 17 · cited 1×
“Among these methods, there are the Canny [9,10] and Sobel [11] algorithms, which detect the edges in an image, and the various corner detection algorithms [12,13], which can provide good repeatability along with localization [14].”
MethodPreparation and Integration of ALHAT Precision Landing Technology for Morpheus Flight Testing
Carson, J. M., Trawny, N., Robertson, E., Roback, V. E., Pierrottet, D. F., Devolites, J. L., Hart, J., & Estes, J. N. (2014) · AIAA SPACE Conference & Exposition · doi:10.2514/6.2014-4313 · Cited by 15
PSO-Based Soft Lunar Landing with Hazard Avoidance: Analysis and Experimentation
D’Ambrosio, A., Carbone, A., Spiller, D., et al. (2021) · Aerospace · doi:10.3390/aerospace8070195 · Cited by 11 · cited 1×
“Among these methods, there are the Canny [9,10] and Sobel [11] algorithms, which detect the edges in an image, and the various corner detection algorithms [12,13], which can provide good repeatability along with localization [14].”
MethodSmall-body segmentation based on morphological features with a u-shaped network architecture
Pugliatti, M., & Maestrini, M. (2022) · Journal of Spacecraft and Rockets · doi:10.2514/1.a35447 · Cited by 8 · cited 1×
“Based on a deep convolutional encoder/decoder (DCE) structure, this architecture is known for its versatility and high accuracy in various applications [20–22].”
MethodWei, R., Jiang, J., Ruan, X., et al. (2018) · Earth Moon and Planets · doi:10.1007/s11038-018-9516-2 · Cited by 5 · cited 1×
“D) laser-elevation-image-based [5,6].”
Guidance navigation and control for Chang’E-4 lander
Zhang, H., YiFeng, G. U. A. N., Cheng, M., et al. (2019) · doi:10.1360/sst-2019-0109 · Cited by 4 · cited 1×
“The following Chang’e 4 and 5 landers used a similar design to Chang’e 3 [7,8], confirming the robustness of the design.”
MethodInfluentialRealistic visualization of solar system small bodies using Blender ray tracing software
Penttilä, A., Palos, M. F., & Kohout, T. (2024) · doi:10.5194/epsc2021-791 · Cited by 3 · cited 1×
“This gap can be mitigated, as shown in [34].”
BackgroundISS Monocular Depth Estimation Via Vision Transformer
Ghilardi, L., Scorsoglio, A., & Furfaro, R. (2023) · Studies in Computational Intelligence · doi:10.1007/978-3-031-25755-1_11 · Cited by 3
Self-citation
Research works printed but not linked: 3
- Ghilardi, L., D’Ambrosio, A., Scorsoglio, A., Furfaro, R., and Curti, F. (February, January 31). Image-based lunar landing hazard detection via deep learning. Proceedings of the 31st AAS/AIAA Space Flight Mechanics Meeting, Virtual.
- Scorsoglio, A., D’Ambrosio, A., Ghilardi, L., Furfaro, R., Gaudet, B., Linares, R., and Curti, F. (2020, January 9–13). Safe Lunar landing via images: A Reinforcement Meta-Learning application to autonomous hazard avoidance and landing. Proceedings of the 2020 AAS/AIAA Astrodynamics Specialist Conference, Virtual.
- Boncelet, C. (2009). The Essential Guide to Image Processing, Elsevier.
Cited by
Posada, D., & Henderson, T. A. (2024) · Aerospace · doi:10.3390/aerospace11050351 · Cited by 5
Real-Time Stochastic Terrain Mapping and Processing for Autonomous Safe Landing
Tomita, K., & Ho, K. (2025) · Journal of Spacecraft and Rockets · doi:10.2514/1.a36253 · Cited by 7
Vision-Based Movement Primitives for Lunar Hazard Avoidance
Cloud, J. M., Beksi, W. J., & Schuler, J. M. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11128448 · Cited by 3
“Additionally, vision transformer models [25], [12] are emerging as powerful tools for delineating surface hazards from disparate imagery.”
Instance Segmentation-Based Hazard Detection with Lunar South Pole Lighting
Cloud, J. M., Buckles, B. C., Muller, T. J., Beksi, W. J., & Schuler, J. M. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11127967 · Cited by 3
“More recently, Ghilardi and Furfaro [21] proposed a ViT-based approach for lunar hazard detection under low illumination conditions at the LSP. Trained on 125,000 simulated low-light RGB images, this model outperforms traditional CNNs in extracting terrain features and identifying hazardous areas…”
Scorsoglio, A., Gaudet, B., Ghilardi, L., & Furfaro, R. (2025) · Neural Computing and Applications · doi:10.1007/s00521-025-11274-7 · Cited by 2
“Finally, in [59], the authors present the capabilities of neural network models in low illumination conditions.”
Self-citation
Metadata
- Authors
- Ghilardi, L., & Furfaro, R.
- Affiliations
- University of Arizona (US)
- Corresponding author
- Furfaro, Roberto
- Venue
- Sensors
- Publisher
- MDPI AG
- Published
- 2023-09-13
- Volume
- 23
- Issue
- 18
- Pages
- 7844
- ISSN
- 1424-8220 (ISSN-L)
- PMC
- PMC10535458
- OpenAlex
- W4386713326
- Cited by
- 14
- References
- 36
- Citations by year
- 2026: 6 · 2025: 7 · 2024: 1
- FWCI
- 1.27
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Sensors (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Robotics and Sensor-Based Localization
- Physical Sciences › Engineering › Civil and Structural Engineering › Infrastructure Maintenance and Monitoring
- Keywords
- vision transformer, low-light conditions, terrain feature extraction, lunar south pole, autonomous landing
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 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (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.