Pith. sign in

REVIEW 3 major objections 8 minor 25 references

MOANA: Multi-Radar Dataset for Maritime Odometry and Autonomous Navigation Application

T0 review · 3 major / 8 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 The MOANA dataset claims to be the first maritime multi-radar dataset, pairing long-range X-band and short-range W-band radar to cover berthing, sailing, and docking.

desk verdict A genuinely useful first maritime X-band/W-band radar dataset; the main weakness is thin calibration evidence, not the central claim. read the letter →

arxiv 2412.03887 v4 pith:TNZM7JZC submitted 2024-12-05 cs.RO cs.CV

Hyesu Jang , Wooseong Yang , Hanguen Kim , Dongje Lee , Yongjin Kim , Jinbum Park , Minsoo Jeon , Jaeseong Koh
show 7 more authors
Yejin Kang Minwoo Jung Sangwoo Jung Chng Zhen Hao Wong Yu Hin Chew Yihang Ayoung Kim
This is my paper · ORCID
classification cs.ROcs.CV
keywords DatasetMaritimeRadarLiDARObjectDetectionPlaceRecognitionOdometrySLAM
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper introduces MOANA, a public maritime navigation dataset built around two radar bands with complementary strengths: X-band marine radar for long-range detection over kilometers and W-band imaging radar for high-resolution detection within about 600 meters. The dataset also includes LiDAR, stereo camera images, GNSS poses, and 2D bounding-box labels for one island sequence, across seven sequences in port, island, and seaside environments. The authors' central claim is that this multi-radar configuration is the first of its kind in maritime settings and that fusing the two bands gives more robust navigation than either sensor alone. If the claim holds, researchers gain a shared benchmark for radar odometry, SLAM, place recognition, and object detection in conditions where cameras and LiDAR are unreliable.

What carries the argument

The load-bearing object is the dataset itself, with its synchronized multi-sensor recordings and calibration files. X-band data are Cartesian PNG images with ranges up to 2,498 m (Port) or 3,328 m (Island); W-band data are polar images converted to 1,024 by 1,024 Cartesian images at about 0.175 m per pixel over 600 m; LiDAR comes as point clouds; GNSS poses define the base frame; and one sequence carries 2D bounding-box labels for vessels and buoys. Calibration between the two radars is done by converting W-band polar images to Cartesian and matching prominent pixel-level features, with vertical extrinsics taken from a CAD model; LiDAR-to-W-band alignment uses phase correlation between polar images.

What would settle it

Take any berthing segment where both radars and LiDAR see the same static structures, project the W-band radar and LiDAR into X-band coordinates using the provided extrinsics, and measure the pixel-level alignment of those structures; a systematic offset larger than a few pixels or degrees would break the complementary-fusion claim.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that X-band and W-band radar provide complementary maritime perception: the W-band radar achieves lower odometry error in the near-harbor Near Port sequence (20.35 m absolute trajectory error with the CFEAR method, versus 63.48 m for X-band with LodeStar), while the X-band radar remains usable across all sequences where the W-band radar sees mostly empty water. The dataset is the first to record both radar bands together in a maritime environment, and the benchmark results are offered as evidence that a hybrid approach—short-range high-resolution W-band plus long-range X-band—can support the full range of vessel navigation, from berthing to open-water transits.

Load-bearing premise

The dataset's claimed value for sensor fusion rests on the accuracy of the extrinsic calibration among X-band radar, W-band radar, LiDAR, and cameras, yet the paper reports no residual errors and takes vertical alignments from a CAD model.

Editorial extensions

If this is right

  • Radar odometry and SLAM methods can be benchmarked in maritime settings with both long- and short-range sensors on the same platform.
  • W-band radar can be treated as a LiDAR alternative for near-field maritime perception where fog, salt spray, and corrosion degrade optical sensors.
  • Fusing X-band and W-band radar should maintain navigation continuity in open water and during berthing, where each band alone fails.
  • The labeled Single Island sequence provides a testbed for radar-based object detection under multipath ghosting and occlusion.
  • Sequence overlaps support inter- and intra-sequence place recognition experiments.

Reading between the lines

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

  • The paper does not report calibration residuals or uncertainties, so a user fusing all modalities should independently validate the provided extrinsics before trusting cross-sensor maps.
  • Because the two vessel setups use different W-band radar models (one with horizontal rays, one with downward-tilted rays), algorithms trained or tuned on Port sequences may need adjustment before they transfer to Island sequences.
  • The stated future work of refined GNSS/radar ground truth suggests that odometry evaluations on the current release will carry GNSS drift in regions of signal loss; early adopters may want to build their own reference trajectories for those stretches.
  • One testable extension is to run existing W-band place-recognition methods on the Island sequences to see whether natural features such as trees and rocks give repeatable radar descriptors, which the paper leaves open.
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 / 8 minor

Summary. The paper introduces MOANA, a maritime navigation dataset combining X-band marine radar, W-band scanning radar, LiDAR, stereo cameras, and GNSS, collected across seven sequences in port and island environments in South Korea and Singapore. The dataset includes extrinsic calibration files, GNSS-based trajectories, and 2D bounding-box object labels for one sequence (Single Island). The authors also report radar odometry benchmarks using CFEAR for W-band and LodeStar for X-band, with absolute trajectory error (ATE) results in Table 3. The central contribution claimed is that MOANA is the first multi-radar (X-band plus W-band) maritime dataset, enabling complementary short- and long-range perception for berthing, sailing, and docking scenarios.

Significance. If the calibration and ground truth are trustworthy, MOANA addresses a genuine gap: existing maritime radar datasets such as Pohang Canal provide only X-band data, while W-band radar datasets are predominantly land-based. The combination of a long-range X-band radar and a high-resolution short-range W-band radar, along with LiDAR, cameras, and object labels, could support research in maritime odometry, SLAM, place recognition, and object detection. The paper also provides a public website, ROS-based data publisher, and a per-sensor file structure, which are practical strengths for community adoption. However, the dataset's value hinges on the accuracy of the multi-radar extrinsic calibration and on the reliability of the GNSS-based reference, both of which currently lack quantitative validation. The benchmark results are suggestive but rest on a single W-band sequence and on an X-band method from the same group, so the empirical support for the complementarity claim is limited.

major comments (3)
  1. [§3.2.1] The radar-to-radar extrinsic calibration is a load-bearing component of the dataset's central claim, but the manuscript reports no residuals, no number of feature correspondences, no per-sequence validation, and no uncertainty estimates for the estimated x, y, and yaw parameters. The vertical components are taken from a CAD model. Given that X-band pixels are 2.44 m to 3.25 m and W-band range extends to 600 m, even a small yaw error (e.g., 0.5 degrees) produces a cross-range offset of several meters at the W-band maximum range, which would corrupt the overlap maps in Figure 5 and any future multi-radar fusion. The paper should provide independent validation of the extrinsic transforms, for example by comparing radar-projected features against LiDAR or satellite imagery in overlapping regions, or by reporting translation and rotation residuals from the feature alignment procedure for each sequence.
  2. [§4.1.4 and Table 3] The GNSS ground truth is acknowledged to suffer from signal reception failures and significant positional drift, and the authors explicitly defer a refined ground truth to future work. Table 3 reports ATE values computed against this drifting reference, yet no uncertainty, standard deviation, or per-sequence trajectory quality measures are given. This means the benchmark conclusions—such as 'W-band radar demonstrates lower error in the Near Port sequence' and 'X-band radar exhibits robust performance across all other sequences'—are not quantitatively substantiated. The authors should either release a corrected ground truth, report uncertainty bounds for the ATE values, or clearly label the numeric results as preliminary and support them with qualitative trajectory comparisons.
  3. [§5.1 and Table 3] The odometry benchmark covers only one of seven sequences for W-band radar (Near Port), and the W-band results for the other sequences are omitted because of blank images or tracking loss. The paper's broader claim that 'a complementary sensor configuration enables robust navigation' is therefore not directly demonstrated by the benchmark; it is inferred from the failure modes of each sensor alone. A simple fusion experiment (e.g., switching between W-band and X-band odometry based on range or feature availability) would provide direct evidence for the complementarity claim, or the wording should be softened to indicate that complementarity is a design rationale rather than a demonstrated result.
minor comments (8)
  1. [§4.1] There is a typo in the first sentence: 'GNSS Reciever' should be 'GNSS Receiver.'
  2. [§3.2] The sentence 'The base for the yacht is established using the GNSS data. all the extrinsic calibration data are included in the calibration.' has a capitalization error and reads awkwardly; please revise for clarity.
  3. [Table 3] The table lists ATE values without units; please specify that the values are in meters and add a column or footnote indicating the uncertainty (e.g., standard deviation) for each trajectory.
  4. [§5.2] The X-band odometry method LodeStar is authored by the same group that presents MOANA. To help readers interpret the benchmark, the paper should explicitly note this potential bias and, if possible, include an independent X-band odometry baseline.
  5. [§3.1] The text says 'The first sequence, Port, was captured using a small fishing boat, while the Island sequence was recorded aboard a larger yacht,' but the dataset contains multiple sequences under Port and Island; please clarify that the two hardware configurations correspond to the two regional setups, not to individual sequences.
  6. [§4.2.2] The subsection numbering for the Island sequences is inconsistent: 'Complex Island' is listed as '(iii)' and 'Island Expedition' is also labeled '(iii)'; the latter should be '(iv).'
  7. [§4.1.1] The manual refinement of X-band range estimates using satellite imagery, LiDAR, and W-band radar data is described only briefly; more detail on how this refinement was performed and validated would help users understand the accuracy of the range calibration.
  8. [General] The dataset is currently hosted on a Google Sites page; for long-term accessibility and citability, a DOI or a permanent archive (e.g., Zenodo) would be preferable.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the dataset contribution is self-contained and the benchmark results are reported observations, not fitted inputs or definitional consequences.

full rationale

The paper's central claim is a dataset contribution (first maritime multi-radar X-band/W-band dataset), which is an existence claim supported by the released sensor data, calibration files, and GNSS-based maps; no derivation is fitted to produce the dataset itself. The odometry benchmark in Section 5 uses CFEAR (Adolfsson et al. 2022) and LodeStar (Jang et al. 2024). LodeStar is from overlapping authors, so there is a minor self-citation, but it is not load-bearing: the ATE numbers in Table 3 are presented as observations of algorithm performance on the new data, not as evidence that defines the dataset, and no parameter is fit to make a prediction. The LiDAR/W-band calibration in Section 3.2.2 cites MulRan (Kim et al. 2020) for the phase-correlation approach; this is method reuse, not circularity, because the phase-correlation computation is performed on the new sensor data rather than imported as a result. The radar-to-radar calibration in Section 3.2.1 relies on pixel-level overlap and CAD-model vertical extrinsics without reported residuals; this is an evidence-quality and correctness limitation (and the paper itself flags future ground-truth refinement in Section 4.1.4), but it is not a circular step in the derivation of any claim. The X-band range correction in Section 4.1.1 is a manual refinement using satellite imagery, LiDAR, and W-band data; it is an input preprocessing choice, not a fitted parameter renamed as a prediction. No equation in the paper is equivalent to another by construction, and no central claim reduces to a self-citation chain. The only self-citation concern is the use of LodeStar, which is minor and non-load-bearing, so the circularity score is 1.

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

No free parameters are fitted to data; the central claim is dataset availability, not a fitted law. The axioms are calibration correctness, GNSS ground-truth reliability, annotation reliability, and cross-sequence comparability. No new physical entities are postulated.

assumptions (4)
  • domain assumption Radar extrinsic calibration is accurate enough for fusion.
    W-band/X-band calibration is done by pixel-level feature overlap after polar-to-Cartesian conversion, with vertical extrinsics from a CAD model; no residual error is reported (Section 3.2.1).
  • domain assumption GNSS poses are adequate ground truth for odometry benchmarking.
    Section 4.1.4 reports signal reception failures causing positional drift and says refined ground truth is future work; the Table 3 ATE values are measured against this GNSS base.
  • domain assumption The annotations identify genuine objects rather than radar artifacts.
    Section 4.1.5 uses cross-sensor manual annotation to exclude ghost and multipath returns; annotation errors would propagate into detection benchmarks. Only the Single Island sequence is annotated.
  • domain assumption Two vessel configurations produce comparable cross-sequence data.
    Different vessels, W-band models (RAS6-DEV-A versus RAS6-DEV-X), and radar orientations are used across Port and Island sequences; calibration files are provided but cross-sequence consistency is assumed rather than quantified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MOANA: Multi-Radar Dataset for Maritime Odometry and Autonomous Navigation Application." pith.science (2026). https://pith.science/paper/TNZM7JZC

@misc{pith2026241203887,
  author       = {Pith},
  title        = {Pith review of: MOANA: Multi-Radar Dataset for Maritime Odometry and Autonomous Navigation Application},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TNZM7JZC}},
  note         = {Machine review of arXiv:2412.03887}
}
read the original abstract

Maritime environmental sensing requires overcoming challenges from complex conditions such as harsh weather, platform perturbations, large dynamic objects, and the requirement for long detection ranges. While cameras and LiDAR are commonly used in ground vehicle navigation, their applicability in maritime settings is limited by range constraints and hardware maintenance issues. Radar sensors, however, offer robust long-range detection capabilities and resilience to physical contamination from weather and saline conditions, making it a powerful sensor for maritime navigation. Among various radar types, X-band radar is widely employed for maritime vessel navigation, providing effective long-range detection essential for situational awareness and collision avoidance. Nevertheless, it exhibits limitations during berthing operations where near-field detection is critical. To address this shortcoming, we incorporate W-band radar, which excels in detecting nearby objects with a higher update rate. We present a comprehensive maritime sensor dataset featuring multi-range detection capabilities. This dataset integrates short-range LiDAR data, medium-range W-band radar data, and long-range X-band radar data into a unified framework. Additionally, it includes object labels for oceanic object detection usage, derived from radar and stereo camera images. The dataset comprises seven sequences collected from diverse regions with varying levels of \bl{navigation algorithm} estimation difficulty, ranging from easy to challenging, and includes common locations suitable for global localization tasks. This dataset serves as a valuable resource for advancing research in place recognition, odometry estimation, SLAM, object detection, and dynamic object elimination within maritime environments. Dataset can be found at https://sites.google.com/view/rpmmoana.

Figures

Figures reproduced from arXiv: 2412.03887 by the authors.

Figure 1
Figure 1. Overview of the MOANA dataset. To address the limitations of individual sensor performance in maritime environments, we enhanced the sensor system by integrating complementary sensors to improve both range and resolution. Data acquisition was carried out across diverse scenarios using multiple sensor types, providing a robust dataset with annotated labels for the development of learning-based algorithms. awareness a… view at source ↗
Figure 2
Figure 2. Modeling and real-world capture setups for two distinct configurations. The primary difference between the hardware setups is the orientation of the GNSS receiver: in the port sequence, the vessel’s forward direction aligns with the x-axis, whereas the island sequence is rotated by 90°. Additionally, the W-band radar is positioned on the right side in the port sequence and on the left side in the island sequence. De… view at source ↗
Figure 3
Figure 3. Composition of sensor types and example images. Radar sensors generate 360-degree scanning bird’s-eye view images, while LiDAR provides 360-degree point cloud data. The camera system captures stereo images in the forward direction. known geometries. The initial transformation for the extrinsic calibration was given by the CAD model. 4 MOANA Dataset 4.1 Data Composition As detailed in Section §3.1, our dataset includ… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Data organization and file structure for each sequence of MOANA dataset. GNSS poses are provided as text files. Extrinsic calibration parameters for the camera, LiDAR, and radars are defined relative to the base frame. Sensor data is available with distinguished file f…
Figure 5
Figure 5. Figure 5: Map constructions from GNSS data-based accumulation of the Island sequences. The X-band radar (b) provides full coverage of the dataset region due to its km level range and strong penetration capability; however, it exhibits relatively low resolution and accuracy. In c…
Figure 6
Figure 6. Figure 6: Examples of 2D annotation for the stereo camera, X-band radar, and W-band radar data. (a,b) The bounding boxes of the detected objects (vessels, buoys) are represented as red boxes. (c,d) yellow and gray pixels are from the X-band and W-band radar each, and the boundin…
Figure 7
Figure 7. Figure 7: Ground truth trajectories for all sequences. The Port sequences (b, c) were collected in South Korea, while the Seaside and Island sequences (d, e, f, g, h) were acquired in Singapore. All sequences share slight common locations except for the Island Expedition sequenc…
Figure 8
Figure 8. Figure 8: W-band and X-band radar odometry/mapping results. The Port sequences achieve promising odometry performance due to the presence of continuous features such as walls and bridges. In contrast, the Island sequences, which lack these features, show insufficient performance…
Figure 9
Figure 9. Figure 9: Odometry estimation results using W-band radar for the Single Island sequence. The W-band radar successfully generates reliable odometry and mapping outcomes with existing methods near the berthing area (blue) of the Single Island sequence. However, tracking loss frequ…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 15 canonical work pages

  1. [1]

    abe and A. Milioto and T. R\

    11em plus .33em minus .07em @technote 4000 4000 100 4000 4000 500 `\.=1000 = #1 #1 #1 0pt [0pt][0pt] #1 * \| ** #1 \@IEEEauthorblockNstyle \@IEEEauthorblockAstyle \@IEEEauthordefaulttextstyle \@IEEEauthorblockconfadjspace -0.25em \@IEEEauthorblockNtopspace 0.0ex \@IEEEauthorblockAtopspace 0.0ex \@IEEEauthorblockNinterlinespace 2.6ex \@IEEEauthorblockAinte...

    arXiv 2023

  2. [2]

    IEEE Transactions on robotics 39(2): 1476--1495

    Adolfsson D, Magnusson M, Alhashimi A, Lilienthal AJ and Andreasson H (2022) Lidar-level localization with radar? the cfear approach to accurate, fast, and robust large-scale radar odometry in diverse environments. IEEE Transactions on robotics 39(2): 1476--1495

    work page 2022

  3. [3]

    In: Proc

    Barnes D, Gadd M, Murcutt P, Newman P and Posner I (2020) The oxford radar robotcar dataset: A radar extension to the oxford robotcar dataset. In: Proc. IEEE Intl. Conf. on Robot. and Automat. pp. 6433--6438

    work page 2020

  4. [4]

    In: Proc

    Bovcon B, Muhovi c J, Per s J and Kristan M (2019) The mastr1325 dataset for training deep usv obstacle detection models. In: Proc. IEEE / RSJ Intl. Conf. on Intell. Robots and Sys. pp. 3431--3438

    work page 2019

  5. [5]

    Burnett K, Yoon DJ, Wu Y, Li AZ, Zhang H, Lu S, Qian J, Tseng WK, Lambert A, Leung KY, Schoellig AP and Barfoot TD (2023) Boreas: A multi-season autonomous driving dataset. Intl. J. of Robot. Research 42(1-2): 33--42

    work page 2023

  6. [6]

    IEEE Robot

    Cheng Y, Jiang M, Zhu J and Liu Y (2021 a ) Are we ready for unmanned surface vehicles in inland waterways? the usvinland multisensor dataset and benchmark. IEEE Robot. and Automat. Lett. 6(2): 3964--3970

    work page 2021

  7. [7]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Cheng Y, Zhu J, Jiang M, Fu J, Pang C, Wang P, Sankaran K, Onabola O, Liu Y, Liu D and Bengio Y (2021 b ) Flow: A dataset and benchmark for floating waste detection in inland waters. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 10953--10962

    work page 2021

  8. [8]

    ://arxiv.org/abs/2412.06192

    Choi J, Cho D, Lee G, Kim H, Yang G, Kim J and Cho Y (2024) Polaris dataset: A maritime object detection and tracking dataset in pohang canal. ://arxiv.org/abs/2412.06192

    arXiv 2024

Show all 25 references
  1. [9]

    Chung D, Kim J, Lee C and Kim J (2023) Pohang canal dataset: A multimodal maritime dataset for autonomous navigation in restricted waters. Intl. J. of Robot. Research 42(12): 1104--1114

    2023

  2. [10]

    In: OCEANS 2021: San Diego – Porto

    DeFilippo M, Sacarny M and Robinette P (2021) Robowhaler: A robotic vessel for marine autonomy and dataset collection. In: OCEANS 2021: San Diego – Porto. pp. 1--7. doi:10.23919/OCEANS44145.2021.9705871

    2021

  3. [11]

    (2024) Oord: The oxford offroad radar dataset

    Gadd M, De Martini D, Bartlett O, Murcutt P, Towlson M, Widojo M, Mu s at V, Robinson L, Panagiotaki E, Pramatarov G et al. (2024) Oord: The oxford offroad radar dataset. arXiv preprint arXiv:2403.02845

    2024 arXiv

  4. [12]

    In: Proc

    Geiger A, Lenz P and Urtasun R (2012) Are we ready for autonomous driving? the kitti vision benchmark suite. In: Proc. IEEE Conf. on Comput. Vision and Pattern Recog

    2012

  5. [13]

    IEEE Transactions on Robotics

    Harlow K, Jang H, Barfoot TD, Kim A and Heckman C (2024) A new wave in robotics: Survey on recent mmwave radar applications in robotics. IEEE Transactions on Robotics

    2024

  6. [14]

    IEEE Robotics and Automation Letters

    Jang H, Jung M, Jeon MH and Kim A (2024) Lodestar: Maritime radar descriptor for semi-direct radar odometry. IEEE Robotics and Automation Letters

    2024

  7. [15]

    In: Proc

    Kim G, Park YS, Cho Y, Jeong J and Kim A (2020) Mulran: Multimodal range dataset for urban place recognition. In: Proc. IEEE Intl. Conf. on Robot. and Automat. pp. 6246--6253

    2020

  8. [16]

    In: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA)

    Kim H, Jung M, Noh C, Jung S, Song H, Yang W, Jang H and Kim A (2025) Hercules: Heterogeneous radar dataset in complex urban environment for multi-session radar slam. In: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA). Atlanta

    2025

  9. [17]

    IEEE Transactions on Intelligent Transportation Systems 23(9): 15487--15497

    Lin J, Diekmann P, Framing CE, Zweigel R and Abel D (2022) Maritime environment perception based on deep learning. IEEE Transactions on Intelligent Transportation Systems 23(9): 15487--15497. doi:10.1109/TITS.2022.3140933

    2022

  10. [18]

    doi:https://doi.org/10.1177/02783649231153020

    Nirgudkar S, DeFilippo M, Sacarny M, Benjamin M and Robinette P (2023) Massmind: Massachusetts marine infrared dataset. doi:https://doi.org/10.1177/02783649231153020

    2023 doi

  11. [19]

    IEEE Transactions on Circuits and Systems for Video Technology 29(9): 2720--2732

    Ribeiro R, Cruz G, Matos J and Bernardino A (2017) A data set for airborne maritime surveillance environments. IEEE Transactions on Circuits and Systems for Video Technology 29(9): 2720--2732

    2017

  12. [20]

    In: IEEE International Conference on Robotics and Automation (ICRA)

    Sheeny M, De Pellegrin E, Mukherjee S, Ahrabian A, Wang S and Wallace A (2021) Radiate: A radar dataset for automotive perception in bad weather. In: IEEE International Conference on Robotics and Automation (ICRA). IEEE, pp. 1--7

    2021

  13. [21]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition workshops

    Zhang MM, Choi J, Daniilidis K, Wolf MT and Kanan C (2015) Vais: A dataset for recognizing maritime imagery in the visible and infrared spectrums. In: Proceedings of the IEEE conference on computer vision and pattern recognition workshops. pp. 10--16

    2015

  14. [22]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution isbn journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence...

  15. [23]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  16. [24]

    , " * write output.state after.block = add.period write newline

    ENTRY address archive author booktitle chapter doi edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type url volume year label INTEGERS output.state before.all mid.sentence after.sentence aft...

  17. [25]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

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