The DOI matches its Crossref record; 64 of the 71 references it cites were found in a registry or an index; 16 of the 55 sources its claims cite were checked for support.
Space-Based Global Maritime Surveillance. Part II: Artificial Intelligence and Data Fusion Techniques
Open accessAbstract
Maritime surveillance (MS) is of paramount importance for search and rescue operations, fishery monitoring, pollution control, law enforcement, migration monitoring, and national security policies. Since ground-based radars and automatic identification system (AIS) do not always provide a comprehensive and seamless coverage of the entire maritime domain, the use of space-based sensors is crucial to complement them. We reviewed space-based technologies for MS in the first part of this work, titled Space-based Global Maritime Surveillance. Part I: Satellite Technologies. However, MS systems combining multiple terrestrial and space-based sensors with additional information sources require dedicated artificial intelligence and data fusion techniques for processing raw satellite images and fusing heterogeneous information. The second part of our work focuses on some recent promising artificial intelligence and data fusion techniques for MS using space-based sensors.
Overview
CoChat Summary
TL;DR
This review surveys artificial intelligence, deep learning, and Bayesian/data-fusion techniques for space-based maritime surveillance, covering satellite image segmentation/classification, knowledge discovery from Sat-AIS, anomaly detection, and sum-product-algorithm-based multitarget tracking. It illustrates these methods with use cases on VHR/HR/Sentinel-2/COSMO-SkyMed imagery and AIS/SAR fusion, including an MTG extraction and a Malta-area SAR/AIS tracking scenario.
Key findings
- DNN use cases include Mask R-CNN trained on about 200,000 VHR images (≈1.5 m resolution, 768×768 pixels) from GeoEye, SPOT, Pleiades, and Black Sky for ship detection/feature extraction; U-Net on a synthetic HR dataset obtained by degrading 10,000 VHR SPOT images; ResNet34 for stern/bow classification on Sentinel-2 clipped ships; and an AIS-based vessel classifier with 2 hidden layers (200 and 100 neurons) and 14 output categories.
- Historical Sat-AIS knowledge discovery via maritime traffic graphs (MTGs) is demonstrated with 19,009 AIS tracks, 162,662 detected waypoints, 2,286 waypoint clusters, and a pruned graph of 99 nodes and 401 edges; the method was validated on real AIS datasets from the Iberian Coast and English Channel and in four MARISA operational trials.
- Maritime anomaly detection is formulated using the Ornstein-Uhlenbeck mean-reverting model with generalized likelihood ratio testing and joint Bayesian detection/tracking via hybrid Bernoulli filtering; it can detect deviations from standard routes and “dark” deviations during surveillance gaps, and was applied to a real-world rendezvous detection scenario.
- SPA-based multisensor MTT is presented as flexible and scalable for heterogeneous fusion, Sat-AIS integration, and class-aided tracking; a use case off Malta from 8 September 2018 at 16:30 to 12 September 2018 at 16:47 fused COSMO-SkyMed SAR measurements with AIS messages (AIS interval at least 5 hours) and identified a single target, the vehicles carrier GAIA LEADER, with real and estimated trajectories.
Methods
- Design: review/overview paper, companion to Part I, synthesizing AI/DNN and Bayesian/data-fusion techniques for maritime surveillance using space-based sensors.
- Data/sample: satellite images (VHR ≈1.5 m, HR ≈10 m, Sentinel-2, COSMO-SkyMed SAR), historical Sat-AIS data, AIS messages, and SAR measurements; use cases include training datasets of about 200,000 VHR images and 10,000 degraded VHR images.
- Main analysis: descriptive review and illustrative use cases rather than a controlled comparative experiment; quantitative results are limited mainly to dataset/graph counts and the Malta SAR/AIS tracking example.
Limitations
- The paper does not state an explicit limitations section. It notes Sat-AIS data challenges: erroneous, incomplete, intermittent, or counterfeit information; irregular sampling; coverage gaps from minutes to many hours; and intentional AIS disablement.
- It states limitations of existing MTT algorithms: fusion of heterogeneous information sources is not straightforward, they do not adapt to time-varying model parameters, and their complexity usually does not scale well with relevant system parameters such as the number of sensors.
- Evident from the text: as a review, the use cases are illustrative and not comprehensive; most DNN use cases report no accuracy or performance metrics, and the second HR segmentation use case relies on synthetic data because a real HR dataset was unavailable (the latter is stated in the paper).
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
9 Valid · 8 Not valid · 19 Not checked · 41 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Complex DNNs with multiple hidden layers and neurons can be used for satellite image segmentation and target classification.”
What the sources say
- Could not verifycites [2] [2] I. Goodfellow, Y. Bengio, and A. Courville, 'Deep learning,' 2016, book in preparation for MIT Press. [Online]. Available: http://www.deeplearningbook.org - sourceExperimental — accuracy not yet measured
Our copy of Vivone et al. (2015) is entirely about knowledge-based multi-target ship tracking with HF surface-wave radar using a VS-IMM estimator combined with JPDA and UKF (Abstract p. 2; Index Terms p. 3; Introduction pp. 4-5; Conclusions p. 30). No passage mentions deep neural networks, hidden layers/neurons, satellite image segmentation, or learned target classification; searches on 'deep neural network hidden layers satellite image segmentation', 'neural network machine learning classifier' and 'image segmentation deep learning hidden layers neurons' returned only tracking/clutter passages (e.g. E. Adaptive Filter Module Selection, p. 13). The only satellite reference is a background remark that SAR satellites cannot give continuous coverage (Introduction, p. 3), which neither supports nor contradicts the DNN claim, so the claim cannot be checked against this source. (LLM, conf=0.75)
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.
- Could not verify
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Space-based sensors for Earth observation installed on satellites allow collecting images of very large areas in remote regions of the globe with relatively short latency, and hence are strongly relevant to MS.”
What the sources say
- Could not verifycites [1] [1] G. Soldi, D. Gaglione, N. Forti, A. D. Simone, F. C. Daffin` a, G. Bottini, D. Quattrociocchi, L. M. Millefiori, P. Braca, S. Carniel, P. Willett, A. Iodice, D. Riccio, and A. Farina, 'Space-based global maritime surveillance. Part I: Satellite technologies,' IEEE Aerosp. Electron. Syst. Mag. (Submitted) , 2020. - sourceExperimental — accuracy not yet measured
The introduction (p. 1) discusses radar, sonar, and cameras as sensors for multitarget tracking, but does not address satellite Earth-observation imagery. Searches for satellite imagery, remote-area coverage, and latency found no passage supporting the attributed claim. (LLM, conf=0.76)
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.
- Could not verify
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Besides the most common terrestrial sensors for MS, e.g., the automatic identification system (AIS), X-band radars, and over-the-horizon (OTH) radars, space-based sensor technologies enable persistent monitoring of the maritime domain and ship traffic on a global scale even in remote areas of the Earth.”
What the sources say
- Could not verifycites [1] [1] G. Soldi, D. Gaglione, N. Forti, A. D. Simone, F. C. Daffin` a, G. Bottini, D. Quattrociocchi, L. M. Millefiori, P. Braca, S. Carniel, P. Willett, A. Iodice, D. Riccio, and A. Farina, 'Space-based global maritime surveillance. Part I: Satellite technologies,' IEEE Aerosp. Electron. Syst. Mag. (Submitted) , 2020. - sourceExperimental — accuracy not yet measured
The introduction (p. 1) discusses multitarget tracking for surveillance using sensors such as radar, sonar, and cameras; the conclusion (p. 12) concerns the tracking algorithm. Searches for space-based sensors, maritime ship traffic, and global monitoring found no passage addressing the claimed persistent coverage of remote areas. (LLM, conf=0.76)
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.
- Could not verify
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“An emerging approach to MTT and information fusion - one with flexibility, low complexity and useful scalability - is based on a factor graph and the sum-product algorithm (SPA) [38], [39], [49].”
What the sources say
- Supportedcites [38] [38] F. Meyer, P. Braca, P. Willett, and F. Hlawatsch, 'A scalable algorithm for tracking an unknown number of targets using multiple sensors,' IEEE Trans. Signal Process. , vol. 65, no. 13, pp. 3478-3493, Jul. 2017. - source
The Proposed Method and Other Message Passing Methods (p. 2) describes multisensor multitarget tracking based on a factor graph and belief propagation using the sum-product algorithm. The same section reports low complexity, favorable scaling, and suitability for general nonlinear, non-Gaussian models, supporting the cited description of flexibility. (LLM, conf=0.94)
- Not checked — no full text of the sourcecites [39] [39] F. Meyer, T. Kropfreiter, J. L. Williams, R. A. Lau, F. Hlawatsch, P. Braca, and M. Z. Win, 'Message passing algorithms for scalable multitarget tracking,' Proc. IEEE , vol. 106, no. 2, pp. 221-259, Feb. 2018. - source
- Could not verifycites [49] [49] D. Gaglione, G. Soldi, F. Meyer, F. Hlawatsch, P. Braca, A. Farina, and M. Z. Win, 'Bayesian information fusion and multitarget tracking for maritime situational awareness,' IET Radar Sonar Navi. (in press) , 2020. - sourceExperimental — accuracy not yet measured
Section 8.1 describes vessel tracking by associating detections near predicted positions, but does not discuss a factor graph or the sum-product algorithm. Searches for those terms and for the claimed flexibility, low complexity, and scalability found no passage addressing this approach. (LLM, conf=0.76)
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.
- Supported
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“First, the fusion of heterogeneous information sources is not straightforward.”
What the sources say
- Could not verifycites [47] [47] A. Ponsford and J. Wang, 'A review of high frequency surface wave radar for detection and tracking of ships,' Turk. J. Elec. Eng. & Comp. Sci. , vol. 18, pp. 409-428, May 2010. - sourceExperimental — accuracy not yet measured
Section 8.1 describes associating radar detections into vessel tracks, while Section 5.3 discusses a trade-off in Doppler processing. Neither addresses the difficulty of fusing heterogeneous information sources in MTT algorithms. Searches for “heterogeneous,” “fusion,” and multi-target tracking across different sources found no passage making that point. (LLM, conf=0.76)
- Not checked — no full text of the sourcecites [48] [48] S. Maresca, P. Braca, J. Horstmann, and R. Grasso, 'Maritime surveillance using multiple high-frequency surface-wave radars,' IEEE Trans. Geosci. Remote Sens. , vol. 52, no. 8, pp. 5056-5071, Aug. 2014. - source
plain- 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.
- Could not verify
34 more claims · 7 with their cited sources checked
Create a free account to see all 77 claims and to verify the 2 references still open.
- Probably not a substantial claim. 27
- It may state two separate claims as one. 16
- Borderline: one of the tests for a substantial claim is unclear. 1
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 · cited 2×
“The type of architecture employed in this case is the U-Net, and The third use case is related to the classification between ship stern and bow by means of a ResNet34 [9] architecture.”
BackgroundU-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 3×
“The type of architecture employed in this case is the U-Net, and The third use case is related to the classification between ship stern and bow by means of a ResNet34 [9] architecture.”
MethodHe, K., Gkioxari, G., Dollar, P., & Girshick, R. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.322 · Cited by 29,137
(1996) · Cited by 20,179
(2016) · Cited by 11,679
Estimation with Applications to Tracking and Navigation: Theory, Algorithms and Software
Bar‐Shalom, Y., Li, X. ‐., & Kirubarajan, T. (2002) · doi:10.1002/0471221279 · Cited by 6,782
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2015) · arXiv · doi:10.48550/arxiv.1512.03385 · Cited by 4,196
An algorithm for tracking multiple targets
Reid, D. (1979) · IEEE Transactions on Automatic Control · doi:10.1109/tac.1979.1102177 · Cited by 3,059
Statistical Multisource-Multitarget Information Fusion
Mahler, R. (2007) · Cited by 2,456
Multitarget bayes filtering via first-order multitarget moments
Mahler, R. P. S. (2003) · IEEE Transactions on Aerospace and Electronic Systems · doi:10.1109/taes.2003.1261119 · Cited by 2,235
Research works printed but not linked: 1
- adelson, 1984, Pyramid methods in image processing, RCA Eng, 29, 33
Cited by
Artificial Intelligence for Trusted Autonomous Satellite Operations
Thangavel, K., Sabatini, R., Gardi, A., Ranasinghe, K. C., Hilton, S., Servidia, P. A., & Spiller, D. (2023) · Progress in Aerospace Sciences · doi:10.1016/j.paerosci.2023.100960 · Cited by 144
Nguyen, D., & Fablet, R. (2024) · IEEE Access · doi:10.1109/access.2024.3349957 · Cited by 156
“…time horizons, ranging from a few tens of minutes to tens of hours ahead, is at the core of diverse MS and MSA applications, including but not limited to search and rescue [1], [2], traffic control [3], path planning [4], [5], [6], port congestion avoidance [7], [8], [9], pollution monitoring [10].”
BackgroundSpace-Based Global Maritime Surveillance. Part I: Satellite Technologies
Soldi, G., Gaglione, D., Forti, N., Simone, A. D., Daffinà, F. C., Bottini, G., Quattrociocchi, D., Millefiori, L. M., Braca, P., Carniel, S., et al. (2021) · IEEE Aerospace and Electronic Systems Magazine · doi:10.1109/maes.2021.3070862 · Cited by 98
“The second part of this work [1] focuses on artificial intelligence and”
BackgroundInfluentialSelf-citationZhang, Z., Zhang, L., Wu, J., & Guo, W. (2024) · IEEE Geoscience and Remote Sensing Magazine · doi:10.1109/mgrs.2024.3404506 · Cited by 110
The usefulness of artificial intelligence for safety assessment of different transport modes
Tselentis, D. I., Papadimitriou, E., & van Gelder, P. H. A. J. M. (2023) · Accident Analysis & Prevention · doi:10.1016/j.aap.2023.107034 · Cited by 98
Metadata
- Authors
- Soldi, G., Gaglione, D., Forti, N., Millefiori, L. M., Braca, P., Carniel, S., Simone, A. D., Iodice, A., Riccio, D., Daffina, F. C., Quattrociocchi, D., Bottini, G., Willett, P., & Farina, A.
- Affiliations
- NATO Centre for Maritime Research and Experimentation (IT)
- University of Naples Federico II (IT)
- e GEOS (Italy) (IT)
- Food and Agriculture Organization of the United Nations (IT)
- University of Connecticut (US)
- ORCID
- Venue
- IEEE Aerospace and Electronic Systems Magazine
- Published
- 2021-09-01
- Volume
- 36
- Issue
- 9
- Pages
- 30-42
- ISSN
- 0885-8985 (ISSN-L)
- 1557-959X
- OpenAlex
- W3110443977
- Cited by
- 44
- References
- 71
- Citations by year
- 2026: 6 · 2025: 7 · 2024: 11 · 2023: 6 · 2022: 7 · 2021: 7
- FWCI
- 4.11
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Aerospace and Electronic Systems Magazine (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- CINECA IRIS Institutial research information system (University of Pisa) (submittedVersion)
- Florence Research (University of Florence) (submittedVersion)
- arXiv (Cornell University)
- Funders
- NATO Allied Command Transformation (SAC000A08)
- NIUVT
- Air Force Office of Scientific Research (FA9500-18-1-0463)
- Topics
- Physical Sciences › Engineering › Ocean Engineering › Maritime Navigation and Safety
- Physical Sciences › Engineering › Media Technology › Advanced Image Fusion Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Target Tracking and Data Fusion in Sensor Networks
- Keywords
- maritime surveillance, satellite sensors, satellite imagery, artificial intelligence, data fusion, heterogeneous data fusion, search and rescue, fisheries monitoring, pollution monitoring, migration monitoring, national security, Automatic Identification System
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 and gpt-6-sol).
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Space-Based Global Maritime Surveillance. Part II: Artificial Intelligence and Data Fusion Techniques. https://app.cochat.ai/papers/10.1109/maes.2021.3070884
- The paper
- Soldi, G., Gaglione, D., Forti, N., Millefiori, L. M., Braca, P., Carniel, S., Simone, A. D., Iodice, A., Riccio, D., Daffina, F. C., Quattrociocchi, D., Bottini, G., Willett, P., & Farina, A. (2021). Space-Based Global Maritime Surveillance. Part II: Artificial Intelligence and Data Fusion Techniques. IEEE Aerospace and Electronic Systems Magazine. https://doi.org/10.1109/maes.2021.3070884
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.