Pith. sign in

REVIEW 3 major objections 4 minor 53 references

SeagrassFinder: Deep Learning for Eelgrass Detection and Coverage Estimation in the Wild

T0 review · 3 major / 4 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 Deep learning models, in particular a Vision Transformer fine-tuned on ImageNet features and trained on underwater-enhanced frames, detect eelgrass presence in towed-camera transect video with AUROC above 0.95, and the temporal mean of…

desk verdict The dataset and detection benchmark are worth publishing; the coverage estimator is a simple post-processing idea that the paper itself shows to be degenerate, and the abstract oversells it. read the letter →

arxiv 2412.16147 v2 pith:JJXSIQZ7 submitted 2024-12-20 cs.CV

Jannik Elsä{ss}er , Laura Weihl , Veronika Cheplygina , Lisbeth Tangaa Nielsen This is my paper · ORCID
classification cs.CV
keywords eelgrassdetectionunderwaterimageclassificationtransferlearningVisionTransformerseagrasscoverageestimationenhancementenvironmentalimpactassessmentvideotransectmonitoring
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

The paper aims to replace the slow, subjective manual review of underwater transect videos with an automated deep-learning pipeline for eelgrass (Zostera marina) presence and coverage. It contributes a new dataset of over 8,300 annotated frames from six transect surveys in Copenhagen harbor, tests four transfer-learned architectures, and reports that a Vision Transformer trained on enhanced images reaches AUROC 0.959 on a held-out transect. It also proposes estimating spatial eelgrass coverage as the temporal mean of binary presence predictions over a 10-second sliding window, arguing this can scale monitoring to far larger video volumes than expert annotation allows. The practical stake is cheaper, more repeatable environmental impact assessments, with the model positioned as a supportive tool for marine biologists rather than a replacement for them.

What carries the argument

The load-bearing machinery is transfer learning: ImageNet-pretrained weights are frozen and only two fully-connected layers are fine-tuned, which keeps each model trainable in about 30 minutes on a single GPU. The Vision Transformer's global attention over image patches is the component that most cleanly separates eelgrass from bare seabed under the difficult lighting, blur, and turbidity of the transect videos. A DNN-based underwater image enhancer, DeepWave-Net, is applied before training and testing to widen color contrast, and its use improves accuracy in most comparisons. For coverage, the central object is the temporal mean $TM_r(i)=\frac{1}{r}\sum_{n=0}^{r-1} p_{i+n}$ of binary presence predictions $p$ over a window of $r=30$ frames, intended to convert a stream of frame-level labels into a percentage-like spatial coverage signal along the transect.

What would settle it

Run a controlled tow over a known seabed where the true spatial eelgrass coverage is fixed by manual quadrat sampling and vary the sled speed and camera angle; if the 10-second temporal mean of frame predictions moves while true coverage does not, the coverage estimator is not measuring what it claims. A simpler decisive check: construct a sequence in which dense but small eelgrass patches occur in half the frames and sparse but large patches occur in all frames; the method assigns the same coverage value to both, so any ground-truth scheme that distinguishes them would falsify it as a spatial coverage measure.

Watch

Extended reading notes

Core claim

The central claim is that eelgrass detection in unrestricted underwater footage is a tractable binary classification task for transfer-learned deep networks, with Vision Transformers outperforming ResNet, InceptionNetV3, and DenseNet under consistent evaluation. On the final held-out transect, the best model, ViT++ (trained and tested on DeepWave-Net-enhanced images), achieves 0.902 accuracy, AUROC 0.959, and a calibration error of 0.087, while DenseNet-201 and the unenhanced Vision Transformer both reach AUROC 0.955. The same detector, applied to every 10th frame of new transects, feeds a coverage estimator computed as the rolling mean over 30 predictions (10 seconds of video), which the authors argue captures ecologically meaningful coverage patterns and generalizes to an EIA site roughly 150 km away. They report a Spearman correlation of 0.3169 against expert coverage labels from that site and explicitly caution that the expert labels are sparse, subjective, and often not repeatable, so the result should be interpreted with care.

Load-bearing premise

The whole coverage estimate rests on the assumption that the fraction of frames in a 10-second window in which the classifier says 'eelgrass present' equals the fraction of seafloor actually covered by eelgrass, an equivalence the paper's own examples show does not hold for sparse-versus-dense and small-patch-versus-large-patch scenes.

Editorial extensions

If this is right

  • A single trained Vision Transformer can score every 10th frame of a transect video, turning a 60-minute survey into a geolocated eelgrass presence map at three predictions per second.
  • The same prediction stream yields eelgrass presence by water depth, a standard indicator for environmental impact assessments, without any extra annotation.
  • Underwater image enhancement gives consistent though modest accuracy gains across architectures and should be included when the compute budget allows.
  • The coverage estimate can in principle be produced during the survey itself, allowing marine biologists to monitor transect quality while the vessel is still at sea.
  • Because the classifier is trained on one species, the pipeline is reusable for other submerged aquatic vegetation with similar visual structure once fresh annotations are collected.

Reading between the lines

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

  • The paper leaves the coverage definition open: a frame-presence rate is at best an ordinal abundance index, not a true spatial coverage fraction, so the 0.3169 correlation should be read as coarse alignment rather than validation.
  • The single-species, single-region dataset bounds the 'in the wild' claim; testing on Zostera marina habitats with different turbidity, substrate, and camera systems would be the natural next validation step.
  • Because all models share ImageNet pretraining, the improvement attributed to the Vision Transformer could partly reflect architecture capacity rather than attention; a fairer comparison would vary pretraining source and fine-tuning budget.
  • The rolling window size of 30 frames is chosen from vessel speed rather than from data; comparing window sizes against expert labels would show whether the temporal mean adds value over a simple per-transect proportion of positive frames.
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 / 4 minor

Summary. This manuscript presents SeagrassFinder, a workflow for eelgrass monitoring from underwater transect videos. The authors release an annotated dataset of 8,324 images extracted from six EIA transects in Copenhagen harbor, describe a custom annotation platform, benchmark ResNet, InceptionNetV3, DenseNet, and Vision Transformer classifiers under transfer learning with and without DeepWave-Net enhancement, and report that a Vision Transformer with enhancement reaches AUROC 0.959 on a held-out transect (LYT-9). They then propose a coverage estimator that averages binary presence predictions over a rolling 10-second window, and they compare it with expert coverage labels from a different EIA site, reporting a Spearman correlation of 0.3169. The paper's core detection claim is the high AUROC; the coverage claim is presented as promising preliminary work.

Significance. The paper's strengths are the public dataset (8,324 annotated images with a Zenodo DOI), the transparent reporting of accuracy, AUROC, and calibration error across several architectures, the annotation-quality analysis, and the fact that the coverage estimator is a post-processing average of classifier outputs rather than a regressor fitted to coverage labels, so there is no circularity. If the detection result holds beyond the single test transect, the tool has clear practical value for environmental impact assessments. The coverage estimator, however, is not yet validated as a measure of spatial coverage, and the statistical support offered for it is weaker than the abstract implies.

major comments (3)
  1. [Sect. 4.5] The Spearman correlation of 0.3169, reported with p = 2.2331e-8, is computed on two strongly autocorrelated time series: the prediction series is a rolling mean over a 30-frame window, and the expert labels are smooth segment annotations on video. The standard Spearman test treats observations as independent, so this p-value is not a valid significance level. A block-permutation test or an effective-sample-size adjustment is needed before the abstract's claim that the method 'aligns with expert manual labels' can be supported. Without such a correction, the quantitative validation is insufficient.
  2. [Sect. 2.6] The paper's own Fig. 6 shows that the temporal mean of binary predictions cannot distinguish sparse-but-consistent frames from dense frames, nor small dense patches from large sparse patches. Therefore the proposed quantity estimates the frequency of 'eelgrass present' frames, not the percentage of seabed covered. The manuscript should either rename this quantity and restrict the claims accordingly, or provide a validation that demonstrates ecological equivalence (for example, comparison against spatially meaningful coverage annotations using a segment-level metric). In addition, the window size r=30 is introduced without a sensitivity analysis, despite the text listing sensitivity to window size as a limitation; a sensitivity curve is needed before this can be presented as a coverage estimator.
  3. [Sect. 3.4 / Table 6] The headline detection result (ViT++ AUROC 0.959, Table 6) is computed on a single held-out transect, LYT-9, which comes from the same survey campaign and the same two days as the training data. No confidence intervals are reported, and no external transect is used for the detection test. The paper's claim that the method generalizes 'in the wild' is therefore not yet established. Please report bootstrap confidence intervals for the metrics and, ideally, evaluate on the external EIA transect used in Sect. 4.5 or another independently collected site.
minor comments (4)
  1. [Fig. 6] The caption and text for Fig. 6 refer to a 'temporal mean of ten samples,' but Sect. 2.6 defines the estimator with r=30 samples (10 seconds). Please reconcile this inconsistency.
  2. [Sect. 4.4] The text states that the two best non-enhanced models have calibration errors of 0.084 and 0.086, but Table 6 reports 0.113 for both DenseNet-201 and ViT, and 0.110 and 0.087 for the enhanced versions. Please correct the text to match the table.
  3. [Sect. 4.5] The list of transects used for coverage mapping (LYT-9, LYT-10, LYT-11, LYT-12, LYT-13, LYT-19, and LYT-20) includes transects that are not part of the annotated dataset in Table 2 (LYT-11, LYT-13, LYT-19) and omits LYT-5 and LYT-14; please clarify which videos were processed and whether these additional videos were annotated for coverage or only used for prediction.
  4. [Sect. 2.5] The sentence 'we are the first to deploy Vision Transformers to detect the presence of eelgrass' is a strong novelty claim that is hard to verify from the cited related work; please soften it or support it with a more systematic search.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the classifier is benchmarked on held-out transect data and the coverage estimator is validated against independent expert labels.

full rationale

The paper's two main outputs are an empirical classification benchmark and a coverage estimator. The classification claim (AUROC > 0.95, Table 6) is evaluated on held-out transect LYT-9, with model weights frozen after training on other transects; the test labels are independent of the training labels, so this is not circular. The coverage estimator (Sect. 2.6) is the temporal mean of binary presence predictions, a post-processing operation with no parameters fitted to coverage labels. Its validation (Sect. 4.5) compares this estimator to expert coverage labels from a different EIA site; this is an external benchmark, not a self-referential reduction, even though the Spearman correlation of 0.3169 is weak and the p-value is unreliable due to autocorrelation. The paper explicitly acknowledges limitations of this estimator (Fig. 6, and the caution that expert labels are 'sparse, highly subjective and often not repeatable'), which are validity concerns rather than circularity. The only self-citation (Juodelyte et al., 2024, where Cheplygina is a co-author) appears in Sec. 5.5 as a speculative suggestion about source dataset choice for future work; it is not load-bearing for any central claim. No uniqueness theorem, ansatz, or fitted input is invoked to force the coverage result. Overall, the derivation chain is self-contained: the classifier is independently benchmarked and the coverage estimator is an explicit heuristic validated against external expert labels.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claims rest on a few domain assumptions: that expert labels are a usable ground truth, that temporal frequency of presence detections reflects spatial coverage, and that ImageNet features transfer to underwater imagery. The only hand-set free parameters are the learning rate and the temporal-mean window size; no new physical entities are invented.

free parameters (2)
  • Temporal mean window size (r) = 30 samples (10 s of video)
    Chosen from assumed ship speed of 1 m/s and assumed 1 m^2 image footprint; actual footprint is not measured, and the paper notes sensitivity to this choice.
  • Learning rate = 4.7e-5
    Set identically for all models without a reported sweep; a hand-picked hyperparameter that affects convergence and final accuracy.
assumptions (4)
  • domain assumption Expert annotations on the external validation transect approximate true eelgrass coverage, despite acknowledged subjectivity and non-repeatability.
    Used as ground truth for the Spearman correlation in Section 4.5; the authors themselves caution that expert labels are sparse and subjective.
  • domain assumption Temporal frequency of binary presence predictions over a sliding window is proportional to spatial seagrass coverage.
    Underlies the coverage estimation method in Section 2.6; the paper's Fig. 6 shows this assumption fails for sparse vs. dense patches.
  • domain assumption ImageNet-pretrained weights transfer useful features to underwater eelgrass images.
    The transfer-learning setup freezes ImageNet pretrained layers; the paper notes domain shift from land to underwater is a known challenge.
  • domain assumption The selected training transects, with relatively balanced class distributions, are representative of the range of conditions the system will encounter.
    Transects were chosen for variability and class balance; this may limit generalization to other sites, depths, and seasons.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SeagrassFinder: Deep Learning for Eelgrass Detection and Coverage Estimation in the Wild." pith.science (2026). https://pith.science/paper/JJXSIQZ7

@misc{pith2026241216147,
  author       = {Pith},
  title        = {Pith review of: SeagrassFinder: Deep Learning for Eelgrass Detection and Coverage Estimation in the Wild},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JJXSIQZ7}},
  note         = {Machine review of arXiv:2412.16147}
}
read the original abstract

Seagrass meadows play a crucial role in marine ecosystems, providing benefits such as carbon sequestration, water quality improvement, and habitat provision. Monitoring the distribution and abundance of seagrass is essential for environmental impact assessments and conservation efforts. However, the current manual methods of analyzing underwater video data to assess seagrass coverage are time-consuming and subjective. This work explores the use of deep learning models to automate the process of seagrass detection and coverage estimation from underwater video data. We create a new dataset of over 8,300 annotated underwater images, and subsequently evaluate several deep learning architectures, including ResNet, InceptionNetV3, DenseNet, and Vision Transformer for the task of binary classification on the presence and absence of seagrass by transfer learning. The results demonstrate that deep learning models, particularly Vision Transformers, can achieve high performance in predicting eelgrass presence, with AUROC scores exceeding 0.95 on the final test dataset. The application of underwater image enhancement further improved the models' prediction capabilities. Furthermore, we introduce a novel approach for estimating seagrass coverage from video data, showing promising preliminary results that align with expert manual labels, and indicating potential for consistent and scalable monitoring. The proposed methodology allows for the efficient processing of large volumes of video data, enabling the acquisition of much more detailed information on seagrass distributions in comparison to current manual methods. This information is crucial for environmental impact assessments and monitoring programs, as seagrasses are important indicators of coastal ecosystem health. This project demonstrates the value that deep learning can bring to the field of marine ecology and environmental monitoring.

Figures

Figures reproduced from arXiv: 2412.16147 by the authors.

Figure 1
Figure 1. The SeagrassFinder Project Pipeline: (1) a vessel performs transects surveys by towing a sled with a camera along the seabed and (2) records videos of the underwater environment. The extracted images (3) are then labeled by human annotators (4) and used to train deep neural networks (DNNs) on the task of detecting eelgrass (5). A trained DNN can now be used to replace the time-consuming manual annotation process. En… view at source ↗
Figure 2
Figure 2. (a) - (c) Sample images of areas with no, sparse and dense submerged aquatic vegetation (SAV). (d) A map of the Copenhagen harbor with transect survey lines (blue) and SAV data overlaid. All SAV maps are based on © Copernicus Sentinel-2 data from 2023-06-13. All SAV maps are based on analysis of Sentinel-2 imagery from 2018 (map created with QGIS v.3.30.2). consider 20 transect surveys (marked blue in Fig. 2d) over … view at source ↗
Figure 4
Figure 4. Two images before (left) and after (right) applying underwater image enhancement [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figures from the paper (9 more)
Figure 3
Figure 3. Figure 3: A screenshot of the SeagrassFinder Annotation Platform accessed through a browser window. The labeling interface is simple and intuitive. red channels on the RGB color range. This enables easier differentiation between the green/brown filamentous algae from the eelgras…
Figure 6
Figure 6. Figure 6: Limitations: even though the level of eelgrass abun￾dance is different in the top scenes compared to the bottom scenes, eelgrass coverage estimation based on a temporal mean (TM) of ten samples gives the same result in both examples. We argue that this method can give …
Figure 7
Figure 7. Figure 7: (a) The number of images on log-scale vs. the number of times the image was annotated. (b) The probability of agreement for each number of annotations. Based on the ratio of “ambiguous” to “mistake” anno￾tations we dismiss the presumption that the annotation task was t…
Figure 8
Figure 8. Figure 8: (a) Eelgrass presence: the lines mark transect surveys conducted along planned survey paths, green dots indicate detected eelgrass presence along each transect, yellow dots indicate eelgrass absence. (b) Eelgrass coverage: an estimation of eelgrass coverage in the bent…
Figure 9
Figure 9. Figure 9: Visual Coverage Estimation: (blue) our new approach based on the temporal mean of ViT predictions of a transect video, (orange) human annotated eelgrass coverage. Elsäßer et al.: Preprint submitted to Elsevier Page 10 of 15 [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Cohens kappa scores for all annotators This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 956200. 8. Data Availability The data used to support our findings is …
Figure 11
Figure 11. Figure 11: Mistakes percentage and disagreement percentage for each annotator. A.2. Glossary of Acronyms We provide a glossary with all acronyms used in our work for reference in Tbl. 7. A.3. Screenshot of SeagrassFinder AP We provide a screenshot of the labeling instruction pag…
Figure 13
Figure 13. Figure 13: Eelgrass presence percentage sorted by depth (m). Marint fagdatacenters gældende tekniske anvisninger, 2013. URL https://ecos.au.dk/forskningraadgivning/fagdatacentre/ marint-fagdatacenter/gaeldende-tekniske-anvisninger. Accessed: 2024-12-9. Antoni Burguera. Segmentat…
Figure 12
Figure 12. Figure 12: A screenshot of the start page of the SeagrassFinder AP showing the visual guidance for the labeling instructions. determine seagrass abundance over a range of water depths (Short and Coles, 2001). We present our results in [PITH_FULL_IMAGE:figures/full_fig_p016_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 41 canonical work pages

  1. [1]

    Seagrass classification using unsupervised curriculum learning (ucl)

    Nosheen Abid, Md Kislu Noman, Gy \"o rgy Kov \'a cs, Syed Mohammed Shamsul Islam, Tosin Adewumi, Paul Lavery, Faisal Shafait, and Marcus Liwicki. Seagrass classification using unsupervised curriculum learning (ucl). Ecological Informatics, 83: 0 102804, 2024

    work page 2024

  2. [2]

    Sources of uncertainty in estimation of eelgrass depth limits

    Thorsten JS Balsby, Jacob Carstensen, and Dorte Krause-Jensen. Sources of uncertainty in estimation of eelgrass depth limits. Hydrobiologia, 704: 0 311--323, 2013

    work page 2013

  3. [3]

    Towards visual detection, mapping and quantification of posidonia oceanica using a lightweight auv

    Francisco Bonin-Font, Miquel Massot Campos, and Gabriel Oliver Codina. Towards visual detection, mapping and quantification of posidonia oceanica using a lightweight auv. IFAC-PapersOnLine, 49 0 (23): 0 500--505, 2016

    work page 2016

  4. [4]

    AU Ecoscience - Marint fagdatacenters gældende tekniske anvisninger, 2013

    Annette Bruhn, Signe Høgslund, Karsten Dahl, Dorte Krause-Jensen, and Michael Bo Rasmussen. AU Ecoscience - Marint fagdatacenters gældende tekniske anvisninger, 2013. URL https://ecos.au.dk/forskningraadgivning/fagdatacentre/marint-fagdatacenter/gaeldende-tekniske-anvisninger. Accessed: 2024-12-9

    work page 2013

  5. [5]

    Segmentation through patch classification: A neural network approach to detect posidonia oceanica in underwater images

    Antoni Burguera. Segmentation through patch classification: A neural network approach to detect posidonia oceanica in underwater images. Ecological Informatics, 56: 0 101053, 2020

    work page 2020

  6. [6]

    Seagrass meadows globally as a coupled social-ecological system: Implications for human wellbeing

    Leanne C Cullen-Unsworth, Lina Mtwana Nordlund, Jessica Paddock, Susan Baker, Len J McKenzie, and Richard KF Unsworth. Seagrass meadows globally as a coupled social-ecological system: Implications for human wellbeing. Marine pollution bulletin, 83 0 (2): 0 387--397, 2014

    work page 2014

  7. [7]

    Dhi - we enable a sustainable future for water, October 2023

    DHI . Dhi - we enable a sustainable future for water, October 2023. URL https://www.dhigroup.com. Publication Title: DHI

    work page 2023

  8. [8]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020

    arXiv 2010

Show all 53 references
  1. [9]

    Seagrass ecosystems as a globally significant carbon stock

    James W Fourqurean, Carlos M Duarte, Hilary Kennedy, N \'u ria Marb \`a , Marianne Holmer, Miguel Angel Mateo, Eugenia T Apostolaki, Gary A Kendrick, Dorte Krause-Jensen, Karen J McGlathery, et al. Seagrass ecosystems as a globally significant carbon stock. Nature geoscience, ...

    2012

  2. [10]

    Animal diet in texas seagrass meadows: 13c evidence for the importance of benthic plants

    Brian Fry and Patrick L Parker. Animal diet in texas seagrass meadows: 13c evidence for the importance of benthic plants. Estuarine and Coastal Marine Science, 8 0 (6): 0 499--509, 1979

    1979

  3. [11]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770--778, 2016

    2016

  4. [12]

    Seagrass ecology

    Marten A Hemminga and Carlos M Duarte. Seagrass ecology. Cambridge University Press, 2000

    2000

  5. [13]

    Densely connected convolutional networks

    Gao Huang, Zhuang Liu, Laurens Van Der Maaten, and Kilian Q Weinberger. Densely connected convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4700--4708, 2017

    2017

  6. [14]

    Source matters: Source dataset impact on model robustness in medical imaging

    Dovile Juodelyte, Yucheng Lu, Amelia Jim \'e nez-S \'a nchez, Sabrina Bottazzi, Enzo Ferrante, and Veronika Cheplygina. Source matters: Source dataset impact on model robustness in medical imaging. In International Workshop on Applications of Medical AI, pages 105--115. Springer, 2024

    2024

  7. [15]

    Transfer learning for medical image classification: a literature review

    Hee E Kim, Alejandro Cosa-Linan, Nandhini Santhanam, Mahboubeh Jannesari, Mate E Maros, and Thomas Ganslandt. Transfer learning for medical image classification: a literature review. BMC medical imaging, 22 0 (1): 0 69, 2022

    2022

  8. [16]

    Imagenet classification with deep convolutional neural networks

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems, 25, 2012

    2012

  9. [17]

    Verified uncertainty calibration

    Ananya Kumar, Percy S Liang, and Tengyu Ma. Verified uncertainty calibration. Advances in Neural Information Processing Systems, 32, 2019

    2019

  10. [18]

    The measurement of observer agreement for categorical data

    JR Landis. The measurement of observer agreement for categorical data. Biometrics, 1977

    1977

  11. [19]

    Counting carbon: A survey of factors influencing the emissions of machine learning

    Alexandra Sasha Luccioni and Alex Hernandez-Garcia. Counting carbon: A survey of factors influencing the emissions of machine learning. arXiv preprint arXiv:2302.08476, 2023

    2023 arXiv

  12. [20]

    Metrics reloaded: recommendations for image analysis validation

    Lena Maier-Hein, Annika Reinke, Patrick Godau, Minu D Tizabi, Florian Buettner, Evangelia Christodoulou, Ben Glocker, Fabian Isensee, Jens Kleesiek, Michal Kozubek, et al. Metrics reloaded: recommendations for image analysis validation. Nature methods, 21 0 (2): 0 195--212, 2024

    2024

  13. [21]

    Deep semantic segmentation in an auv for online posidonia oceanica meadows identification

    Miguel Martin-Abadal, Eric Guerrero-Font, Francisco Bonin-Font, and Yolanda Gonzalez-Cid. Deep semantic segmentation in an auv for online posidonia oceanica meadows identification. IEEE Access, 6: 0 60956--60967, 2018

    2018

  14. [22]

    Imaging and classification techniques for seagrass mapping and monitoring: A comprehensive survey

    Md Moniruzzaman, SM Islam, Paul Lavery, Mohammed Bennamoun, and C Peng Lam. Imaging and classification techniques for seagrass mapping and monitoring: A comprehensive survey. arXiv preprint arXiv:1902.11114, 2019 a

    1902 arXiv

  15. [23]

    Faster r-cnn based deep learning for seagrass detection from underwater digital images

    Md Moniruzzaman, Syed Mohammed Shamsul Islam, Paul Lavery, and Mohammed Bennamoun. Faster r-cnn based deep learning for seagrass detection from underwater digital images. In 2019 digital image computing: techniques and applications (DICTA), pages 1--7. IEEE, 2019 b

    2019

  16. [24]

    Multi-species seagrass detection using semi-supervised learning

    Md Kislu Noman, Syed Mohammed Shamsul Islam, Jumana Abu-Khalaf, and Paul Lavery. Multi-species seagrass detection using semi-supervised learning. In 2021 36th International Conference on Image and Vision Computing New Zealand (IVCNZ), pages 1--6. IEEE, 2021

    2021

  17. [25]

    Improving accuracy and efficiency in seagrass detection using state-of-the-art ai techniques

    Md Kislu Noman, Syed Mohammed Shamsul Islam, Jumana Abu-Khalaf, Seyed Mohammad Jafar Jalali, and Paul Lavery. Improving accuracy and efficiency in seagrass detection using state-of-the-art ai techniques. Ecological Informatics, 76: 0 102047, 2023

    2023

  18. [26]

    Submerged aquatic vegetation: seagrasses

    Alyssa B Novak and Frederick T Short. Submerged aquatic vegetation: seagrasses. In Coastal and Marine Environments, pages 277--288. CRC Press, 2020

    2020

  19. [27]

    A global crisis for seagrass ecosystems

    Robert J Orth, Tim JB Carruthers, William C Dennison, Carlos M Duarte, James W Fourqurean, Kenneth L Heck, A Randall Hughes, Gary A Kendrick, W Judson Kenworthy, Suzanne Olyarnik, et al. A global crisis for seagrass ecosystems. Bioscience, 56 0 (12): 0 987--996, 2006

    2006

  20. [28]

    Pytorch: An imperative style, high-performance deep learning library

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32, 2019

    2019

  21. [29]

    Multi-species seagrass detection and classification from underwater images

    Scarlett Raine, Ross Marchant, Peyman Moghadam, Frederic Maire, Brett Kettle, and Brano Kusy. Multi-species seagrass detection and classification from underwater images. In 2020 Digital Image Computing: Techniques and Applications (DICTA), pages 1--8. IEEE, 2020

    2020

  22. [30]

    Image labels are all you need for coarse seagrass segmentation

    Scarlett Raine, Ross Marchant, Brano Kusy, Frederic Maire, and Tobias Fischer. Image labels are all you need for coarse seagrass segmentation. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 5943--5952, 2024

    2024

  23. [31]

    o ller, Jonas J \

    Gereon Reus, Thomas M \"o ller, Jonas J \"a ger, Stewart T Schultz, Claudia Kruschel, Julian Hasenauer, Viviane Wolff, and Klaus Fricke-Neuderth. Looking for seagrass: Deep learning for visual coverage estimation. In 2018 OCEANS-MTS/IEEE Kobe Techno-Oceans (OTO), pages 1--6. I...

    2018

  24. [32]

    Detection of shallow subtidal corals from ikonos satellite and qtc view (50, 200 khz) single-beam sonar data (arabian gulf; dubai, uae)

    Bernhard M Riegl and Samuel J Purkis. Detection of shallow subtidal corals from ikonos satellite and qtc view (50, 200 khz) single-beam sonar data (arabian gulf; dubai, uae). Remote Sensing of Environment, 95 0 (1): 0 96--114, 2005

    2005

  25. [33]

    Field data sets for seagrass biophysical properties for the eastern banks, moreton bay, australia, 2004--2014

    Chris M Roelfsema, Eva M Kovacs, and Stuart R Phinn. Field data sets for seagrass biophysical properties for the eastern banks, moreton bay, australia, 2004--2014. Scientific Data, 2 0 (1): 0 1--6, 2015

    2004

  26. [34]

    Autonomous boundary inspection of posidonia oceanica meadows using an underwater robot

    Francesco Ruscio, Riccardo Costanzi, Nuno Gracias, Josep Quintana, and Rafael Garcia. Autonomous boundary inspection of posidonia oceanica meadows using an underwater robot. Ocean Engineering, 283: 0 114988, 2023

    2023

  27. [35]

    Imagenet large scale visual recognition challenge

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, et al. Imagenet large scale visual recognition challenge. International journal of computer vision, 115: 0 211--252, 2015

    2015

  28. [36]

    Seagrassdetect: A novel method for the detection of seagrass from unlabelled underwater videos

    Sayantan Sengupta, Bjarne Kj r Ersb ll, and Anders Stockmarr. Seagrassdetect: A novel method for the detection of seagrass from unlabelled underwater videos. Ecological Informatics, 57: 0 101083, 2020

    2020

  29. [37]

    Wavelength-based attributed deep neural network for underwater image restoration

    Prasen Sharma, Ira Bisht, and Arijit Sur. Wavelength-based attributed deep neural network for underwater image restoration. ACM Transactions on Multimedia Computing, Communications and Applications, 19 0 (1): 0 1--23, 2023

    2023

  30. [38]

    Global seagrass research methods

    Frederick T Short and Robert G Coles. Global seagrass research methods. Elsevier, 2001

    2001

  31. [39]

    Seagrassnet manual for scientific monitoring of seagrass habitat

    Frederick T Short, Robert G Coles, and CA Short. Seagrassnet manual for scientific monitoring of seagrass habitat. 2015

    2015

  32. [40]

    An overview of the tesseract ocr engine

    Ray Smith. An overview of the tesseract ocr engine. In Ninth international conference on document analysis and recognition (ICDAR 2007), volume 2, pages 629--633. IEEE, 2007

    2007

  33. [41]

    Rethinking the inception architecture for computer vision

    Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jon Shlens, and Zbigniew Wojna. Rethinking the inception architecture for computer vision. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2818--2826, 2016

    2016

  34. [42]

    Large-scale benchmarking and boosting transfer learning for medical image analysis

    Mohammad Reza Hosseinzadeh Taher, Fatemeh Haghighi, Michael B Gotway, and Jianming Liang. Large-scale benchmarking and boosting transfer learning for medical image analysis. Medical Image Analysis, 102: 0 103487, 2025

    2025

  35. [43]

    Comparison of vision transformers and convolutional neural networks in medical image analysis: a systematic review

    Satoshi Takahashi, Yusuke Sakaguchi, Nobuji Kouno, Ken Takasawa, Kenichi Ishizu, Yu Akagi, Rina Aoyama, Naoki Teraya, Amina Bolatkan, Norio Shinkai, et al. Comparison of vision transformers and convolutional neural networks in medical image analysis: a systematic review. Journ...

    2024

  36. [44]

    A survey on deep transfer learning

    Chuanqi Tan, Fuchun Sun, Tao Kong, Wenchang Zhang, Chao Yang, and Chunfang Liu. A survey on deep transfer learning. In Artificial Neural Networks and Machine Learning--ICANN 2018: 27th International Conference on Artificial Neural Networks, Rhodes, Greece, October 4-7, 2018, P...

    2018

  37. [45]

    Training data-efficient image transformers & distillation through attention

    Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, and Herv \'e J \'e gou. Training data-efficient image transformers & distillation through attention. In International conference on machine learning, pages 10347--10357. PMLR, 2021

    2021

  38. [46]

    Global challenges for seagrass conservation

    Richard KF Unsworth, Len J McKenzie, Catherine J Collier, Leanne C Cullen-Unsworth, Carlos M Duarte, Johan S Ekl \"o f, Jessie C Jarvis, Benjamin L Jones, and Lina M Nordlund. Global challenges for seagrass conservation. Ambio, 48: 0 801--815, 2019 a

    2019

  39. [47]

    Seagrass meadows support global fisheries production

    Richard KF Unsworth, Lina Mtwana Nordlund, and Leanne C Cullen-Unsworth. Seagrass meadows support global fisheries production. Conservation Letters, 12 0 (1): 0 e12566, 2019 b

    2019

  40. [48]

    Opportunities for seagrass research derived from remote sensing: A review of current methods

    Bijeesh Kozhikkodan Veettil, Raymond D Ward, Mariana Do Amaral Camara Lima, Milica Stankovic, Pham Ngoc Hoai, and Ngo Xuan Quang. Opportunities for seagrass research derived from remote sensing: A review of current methods. Ecological Indicators, 117: 0 106560, 2020

    2020

  41. [49]

    Real-time and embedded compact deep neural networks for seagrass monitoring

    Jiangtao Wang, Baihua Li, Yang Zhou, Qinggang Meng, Sante Francesco Rende, and Emanuele Rocco. Real-time and embedded compact deep neural networks for seagrass monitoring. In 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC), pages 3570--3575. IEEE, 2020 a

    2020

  42. [50]

    Compact and fast underwater segmentation network for autonomous underwater vehicles

    Jiangtao Wang, Baihua Li, Yang Zhou, Emanuele Rocco, and Qinggang Meng. Compact and fast underwater segmentation network for autonomous underwater vehicles. In Proceedings of the Asian Conference on Computer Vision, 2020 b

    2020

  43. [51]

    Accelerating loss of seagrasses across the globe threatens coastal ecosystems

    Michelle Waycott, Carlos M Duarte, Tim JB Carruthers, Robert J Orth, William C Dennison, Suzanne Olyarnik, Ainsley Calladine, James W Fourqurean, Kenneth L Heck Jr, A Randall Hughes, et al. Accelerating loss of seagrasses across the globe threatens coastal ecosystems. Proceedi...

    2009

  44. [52]

    A closer look at seagrass meadows: Semantic segmentation for visual coverage estimation

    Franz Weidmann, Jonas J \"a ger, Gereon Reus, Stewart T Schultz, Claudia Kruschel, Viviane Wolff, and Klaus Fricke-Neuderth. A closer look at seagrass meadows: Semantic segmentation for visual coverage estimation. In OCEANS 2019-Marseille, pages 1--6. IEEE, 2019

    2019

  45. [53]

    A review of convolutional neural networks in computer vision

    Xia Zhao, Limin Wang, Yufei Zhang, Xuming Han, Muhammet Deveci, and Milan Parmar. A review of convolutional neural networks in computer vision. Artificial Intelligence Review, 57 0 (4): 0 99, 2024

    2024

Pith tools

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