Pith. sign in

REVIEW 3 major objections 6 minor 42 references

Automated Detection of Salvin's Albatrosses: Improving Deep Learning Tools for Aerial Wildlife Surveys

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Fine-tuning a general bird detector with stronger augmentations lifts F1 on Salvin's albatross surveys from 0.6576 to 0.7504.

desk verdict A modest but honest applied transfer-learning study: fine-tuning BirdDetector with stronger augmentations improves F1 on a new albatross dataset, but permissive IoU matching and missing error bars make the exact gain softer than reported. read the letter →

arxiv 2505.10737 v1 pith:KEWR3C5C submitted 2025-05-15 cs.CV

Mitchell Rogers , Theo Thompson , Isla Duporge , Johannes Fischer , Klemens Pütz , Thomas Mattern , Bing Xue , Mengjie Zhang This is my paper · ORCID
classification cs.CV
keywords Salvin'salbatrosswildlifemonitoringUAVsurveysobjectdetectionBirdDetectorfine-tuningdataaugmentationslicing-aidedhyper-inference
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 asks whether a general-purpose avian detector can be turned into a reliable counter of Salvin's albatrosses on drone imagery from the remote Bounty Islands. Using leave-one-island-out evaluation, it shows that fine-tuning the model on annotated tiles from seven islands raises average F1 from 0.6576 for the zero-shot baseline with slicing inference to 0.7045 with standard fine-tuning, and to 0.7504 when training adds stronger augmentations. The authors argue that this is enough to substantially accelerate manual population counts, while noting that false positives from rocks, penguins, and seals, plus missed birds, remain. The significance is practical: a threatened seabird nesting on almost inaccessible islands can be surveyed from drones with a model that transfers to an unseen island without per-island retraining.

What carries the argument

The load-bearing object is BirdDetector, a general avian detector built on a RetinaNet with a ResNet-50 backbone. The authors keep its architecture and default hyperparameters, change test-time inference to Slicing-Aided Hyper-Inference, which cuts images into overlapping 1000 by 1000 pixel patches and merges detections, and replace the original augmentation set with one that adds HSV shifts, random flips, and random crops of 700 to 1200 pixels resized to 1000 by 1000. Ground truth is supplied as 50 by 50 pixel pseudo-bounding boxes centred on manual point annotations from the drone survey. These components work together: the pretrained detector supplies transferable bird features, slicing inference improves small-object recall, and the stronger augmentation set is what pushes the fine-tuned model past the zero-shot baseline.

What would settle it

Re-annotate one held-out island with precise bounding boxes around every visible albatross, run the best fine-tuned model on that island, and recompute F1 at intersection-over-union thresholds from 0.1 to 0.5. If the score collapses as the threshold rises, the reported improvement is largely an artefact of loose box matching rather than real localisation quality; a simpler check is comparing automated counts island by island with independent manual counts on full orthomosaics and seeing whether false positives and false negatives balance.

Watch

Extended reading notes

Core claim

The central claim is that fine-tuning with target-domain annotations and stronger data augmentation markedly improves detection accuracy over zero-shot inference, and that the improvement holds across eight held-out islands. In the best configuration, average F1 reaches 0.7504, versus 0.6576 for the zero-shot model with slicing-aided inference and 0.5018 for zero-shot without it. Detections are scored against pseudo-bounding boxes made by centring 50 by 50 pixel squares on manual point annotations, with a deliberately low intersection-over-union threshold of 0.1 to absorb annotation offset. The paper interprets the results as evidence that stronger augmentation simulates variation in flight altitude and lighting, improving generalisation to unseen islands, and that overlapping-tile inference will be preferable for future whole-island counts.

Load-bearing premise

The evaluation treats manual point annotations, expanded to 50 by 50 pixel boxes and matched at an intersection-over-union threshold of only 0.1, as ground truth; if annotations miss birds, are systematically offset, or the permissive threshold credits detections that are not actually on birds, the reported F1 values overstate detection quality.

Editorial extensions

If this is right

  • On a held-out island, the strongest fine-tuned configuration reaches average F1 of 0.7504 versus 0.6576 for zero-shot with SAHI, so fine-tuning transfers across islands without per-island training.
  • Stronger augmentations usually raise precision at the cost of recall, so the augmentation choice should be tuned to whether a survey prioritises avoiding false alarms or avoiding missed birds.
  • Slicing-aided inference helps the zero-shot model substantially (0.6576 versus 0.5018), and the paper expects overlapping-tile inference on full orthomosaics to reduce double counting in whole-island population counts.
  • With averages of 29.25 percent false positives and 19.3 percent missed birds, the model partially automates the count workflow rather than fully replacing it.
  • Retraining on all eight islands instead of leaving one out is expected to improve performance beyond the cross-validation estimates.

Reading between the lines

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

  • Beyond the paper, the loose IoU threshold of 0.1 means F1 measures whether a detection lands near a bird, not how precisely it locates one; a counting application could report count error directly.
  • Beyond the paper, the same manual point annotations could support a density-map counting baseline, which would test whether bounding-box detection is even necessary for accurate population estimates.
  • Beyond the paper, the many false positives from penguins and seals suggest that a two-stage pipeline that first finds all birds and then separates species could improve precision more than further augmentation tuning.
  • Beyond the paper, applying the best model to a colony with different lighting or substrate would directly test the paper's hint that performance depends on visual similarity to the training islands.
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 / 6 minor

Summary. This manuscript evaluates the general-purpose BirdDetector model for detecting Salvin's albatrosses in drone orthomosaics of eight Bounty Islands islets. Point annotations from a prior survey are converted to 50x50-pixel pseudo-bounding boxes, yielding 571 supertiles containing 66,635 annotated birds. Four configurations are compared under leave-one-island-out cross-validation: zero-shot without SAHI, zero-shot with SAHI, fine-tuning with the original augmentation schedule, and fine-tuning with stronger augmentations. Detection quality is measured by F1 using an IoU threshold of 0.1 and a confidence threshold of 0.1. The headline result is that the strongest configuration improves average F1 from 0.6576 (zero-shot with SAHI) to 0.7504, and the paper concludes that target-domain fine-tuning and stronger augmentation lead to marked improvements in detection accuracy.

Significance. If the claimed effect is robust, the practical contribution is real: the paper demonstrates a workflow for adapting a general bird detector to a dense, remote seabird colony with modest annotation effort, and it provides per-island generalization results that ecologists can use for monitoring. The evaluation is grounded by the use of an externally pretrained model (BirdDetector), held-out-island evaluation, fixed thresholds that were not tuned on a validation set, and transparent reporting of per-island F1 scores rather than a single pooled number. The main limitations are the permissive matching criterion and the absence of variance information, which I discuss below; neither issue reflects circularity, because the test islands and the pretrained weights provide independent grounding for the empirical comparison.

major comments (3)
  1. [Section 3.2, Table 1] The F1 score is computed with IoU threshold 0.1 and confidence threshold 0.1 against 50x50-pixel pseudo-boxes. As the paper acknowledges, the low IoU threshold was chosen because the pseudo-boxes are inaccurate, but at IoU 0.1 a predicted box of the same size can be offset by roughly 40 pixels and still be counted as a true positive. The reported F1 therefore mainly measures whether a detection falls in the broad neighborhood of a manual point rather than whether the predicted box localizes the bird. Because the manuscript's central claim is a marked improvement in detection accuracy, the metric definition is load-bearing. I ask for a sensitivity check at stricter IoU thresholds (e.g., 0.3 and 0.5) with precision and recall reported separately, so the reader can see whether the ordering in Table 1 survives when localization is required.
  2. [Section 2, Section 3.2] The ground truth is a single set of point annotations converted to fixed 50x50 boxes, and the paper states that annotators sometimes offset points and that the selected set was chosen by visual inspection for comprehensiveness. This makes the pseudo-boxes a noisy proxy for true bird extents, and inter-observer variability is acknowledged but not quantified. Because all variants are evaluated against the same noisy labels, the relative ordering of methods may be robust, but the absolute F1 values and the statement that 29.25% of detections are false positives are optimistic, or at least unverified. Please quantify annotation uncertainty (e.g., annotator agreement on a subset of tiles) or discuss how label noise affects the main comparisons.
  3. [Section 3.2, Table 1] Each leave-one-island-out fold appears to be run once; no random seeds, repeated runs, confidence intervals, or significance tests are reported. The headline gain from 0.7045 (fine-tuned) to 0.7504 (fine-tuned with stronger augmentations) is an average over eight islands, and island-level differences can be small or, for Tunnel Island with baseline fine-tuning, negative relative to zero-shot. Without variance estimates, the marked-improvement claim is not distinguishable from training stochasticity. Please provide results over multiple seeds with means and standard deviations, or a paired statistical test over the eight islands.
minor comments (6)
  1. [Section 1] The text contains the typo 'UA Vs' where 'UAVs' is intended; this appears twice in the introduction.
  2. [Section 3.1] The 'stronger augmentation' condition is not fully specified: the ranges for brightness, contrast, and HSV adjustments, and the probabilities of the individual transformations, are omitted, which makes this key training configuration hard to reproduce. Please add these details or a pointer to released code.
  3. [Section 5] The sentence 'an average of 29.25% of the detections were false positives' is ambiguous: it should state whether this is the mean of per-island false-positive rates or a pooled fraction over all detections.
  4. [Figure 2] The caption says the colored boxes correspond to true positives, false positives, and false negatives, but if the figure is viewed in grayscale, the green/red/blue distinction may be lost; consider adding symbols or a separate legend.
  5. [Section 4] The phrase 'significantly more false positives' uses 'significantly' without a statistical test; 'substantially' would be more accurate unless a test is reported.
  6. [General] The manuscript does not state whether the code, trained weights, or annotations will be made available; an availability statement would improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the central empirical claim is grounded in held-out island evaluation and an externally pre-trained detector, with no fitted parameter recycled as a prediction.

full rationale

This paper reports an empirical comparison of detection models under a fixed evaluation protocol; there is no derivation chain in which a prediction is constructed from its own inputs. The pre-trained BirdDetector weights come from external work [40], the fine-tuned variants are trained on seven islands and evaluated on the eighth via leave-one-island-out cross-validation, and the F1 metric is computed against manual point annotations converted to fixed 50×50-pixel pseudo-boxes. These pseudo-boxes are ground-truth inputs, not outputs of the model, so the comparison is not self-referential. The low IoU threshold (0.1) and low confidence threshold (0.1) are fixed, not optimized on the held-out islands, and are applied identically to zero-shot and fine-tuned settings; this is a measurement-choice limitation noted by the authors, not a circular step. Self-citations [30,31,35] supply data provenance and prior context rather than the load-bearing claim that fine-tuning improves detection, which is supported by held-out F1 values. No fitted parameter or cited uniqueness theorem is recycled as a prediction.

Assumptions & free parameters 6 free parameters · 6 assumptions · 0 invented entities

The central claim depends on hand-chosen evaluation thresholds and on treating point annotations converted to 50x50 boxes as valid ground truth. No new physical or model entities are introduced, and no fitted constants are disguised as predictions.

free parameters (6)
  • Pseudo bounding box size = 50 x 50 pixels
    Chosen in Section 2 to match approximate albatross size; all ground-truth boxes and evaluation matches derive from this size.
  • IoU matching threshold = 0.1
    Section 3.2: a very low intersection-over-union threshold is used to tolerate inaccurate boxes from point annotations.
  • Confidence threshold = 0.1
    Section 3.2: a low confidence threshold is used because many predicted boxes have low scores; this trades precision against recall.
  • Non-maximum suppression threshold = 0.05
    Section 3.2: fixed NMS threshold affects how overlapping detections are merged.
  • Augmentation crop settings = probability 0.8, width 700-1200 px, aspect ratio 0.8-1.2
    Section 3.1: stronger augmentation crops a single bounding box region with random width and aspect ratio, then resizes to 1000x1000.
  • Fine-tuning schedule = 30 epochs, learning rate 0.0001
    Section 3.2: fine-tuning hyperparameters inherited from BirdDetector; the paper does not report a sweep.
assumptions (6)
  • domain assumption Manual point annotations from multiple annotators, after selecting the most comprehensive set per island, are accurate enough to serve as ground truth for training and evaluation.
    Section 2 states that one set of annotations was selected per island by visual inspection and that random tiles were checked for errors, but the accuracy of these labels is not quantified.
  • ad hoc to paper A 50x50 pixel square centered on each annotated point approximates the true extent of a Salvin's albatross in the orthomosaic.
    Section 2 says the box size was selected to match the approximate size of albatrosses; no validation of box fit against the birds is presented.
  • domain assumption Evaluation on supertiles containing at least one annotation is representative of detection performance in the intended counting workflow.
    Section 2 retains only tiles with at least one annotation, so empty tiles where false positives would be counted as birds are excluded from the F1 calculation.
  • domain assumption The pretrained BirdDetector weights from Weinstein et al. [40] provide a suitable initialization for this new domain.
    Section 3 uses BirdDetector as the starting point; the paper does not compare with training from scratch or other backbones.
  • ad hoc to paper F1 computed at an IoU threshold of 0.1 and a confidence threshold of 0.1 is a meaningful measure of detection accuracy for this task.
    Section 3.2 sets these thresholds to account for annotation noise, but does not show how results change with more or less permissive thresholds.
  • domain assumption Leave-one-island-out cross-validation estimates how the model will perform on a newly imaged island.
    Section 3.2 uses LOIOCV; this assumes island-level visual conditions in the held-out set resemble those in the training set.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Automated Detection of Salvin's Albatrosses: Improving Deep Learning Tools for Aerial Wildlife Surveys." pith.science (2026). https://pith.science/paper/KEWR3C5C

@misc{pith2026250510737,
  author       = {Pith},
  title        = {Pith review of: Automated Detection of Salvin's Albatrosses: Improving Deep Learning Tools for Aerial Wildlife Surveys},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KEWR3C5C}},
  note         = {Machine review of arXiv:2505.10737}
}
read the original abstract

Recent advancements in deep learning and aerial imaging have transformed wildlife monitoring, enabling researchers to survey wildlife populations at unprecedented scales. Unmanned Aerial Vehicles (UAVs) provide a cost-effective means of capturing high-resolution imagery, particularly for monitoring densely populated seabird colonies. In this study, we assess the performance of a general-purpose avian detection model, BirdDetector, in estimating the breeding population of Salvin's albatross (Thalassarche salvini) on the Bounty Islands, New Zealand. Using drone-derived imagery, we evaluate the model's effectiveness in both zero-shot and fine-tuned settings, incorporating enhanced inference techniques and stronger augmentation methods. Our findings indicate that while applying the model in a zero-shot setting offers a strong baseline, fine-tuning with annotations from the target domain and stronger image augmentation leads to marked improvements in detection accuracy. These results highlight the potential of leveraging pre-trained deep-learning models for species-specific monitoring in remote and challenging environments.

Figures

Figures reproduced from arXiv: 2505.10737 by the authors.

Figure 1
Figure 1. Location of the Bounty Islands relative to New Zealand, and example bounding boxes of the Salvin’s Albatross. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Example predictions for the zero-shot and improved fine-tuned experiments. The coloured boxes correspond to the true positives [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 39 canonical work pages

  1. [1]

    Abraham and Katrin Berkenbusch

    Edward R. Abraham and Katrin Berkenbusch. Preparation of data for protected species capture estimation, updated to 2017–18. Technical Report 234, New Zealand Ministry for Primary Industries, 2019. 2

    work page 2017

  2. [2]

    Slicing aided hyper inference and fine-tuning for small object detection

    Fatih Cagatay Akyon, Sinan Onur Altinuc, and Alptekin Temizel. Slicing aided hyper inference and fine-tuning for small object detection. In 2022 IEEE International Confer- ence on Image Processing (ICIP), pages 966–970, 2022. 2, 4

    work page 2022

  3. [3]

    CountGD: Multi-modal open-world counting

    Niki Amini-Naieni, Tengda Han, and Andrew Zisserman. CountGD: Multi-modal open-world counting. In Advances in Neural Information Processing Systems , pages 48810– 48837. Curran Associates, Inc., 2024. 5

    work page 2024

  4. [4]

    Barber-Meyer, Gerald L

    Shannon M. Barber-Meyer, Gerald L. Kooyman, and Paul J. Ponganis. Estimating the relative abundance of emperor pen- guins at inaccessible colonies using satellite imagery. Polar Biology, 30(12):1565–1570, 2007. 1

    work page 2007

  5. [5]

    Efficient pipeline for camera trap image review

    Sara Beery, Dan Morris, and Siyu Yang. Efficient pipeline for camera trap image review. arXiv preprint arXiv:1907.06772, 2019. 5

    arXiv 1907

  6. [6]

    Fretwell, Geoffrey French, and Michal Mackiewicz

    Ellen Bowler, Peter T. Fretwell, Geoffrey French, and Michal Mackiewicz. Using deep learning to count albatrosses from space: Assessing results in light of ground truth uncertainty. Remote Sensing, 12(12), 2020. 1, 2, 4

    work page 2020

  7. [7]

    Brack, Andreas Kindel, and Luiz Flamarion B

    Ismael V . Brack, Andreas Kindel, and Luiz Flamarion B. Oliveira. Detection errors in wildlife abundance estimates 5 from unmanned aerial systems (UAS) surveys: Synthesis, solutions, and challenges. Methods in Ecology and Evolu- tion, 9(8):1864–1873, 2018. 4

    work page 2018

  8. [8]

    Appli- cations for deep learning in ecology

    Sylvain Christin, ´Eric Hervet, and Nicolas Lecomte. Appli- cations for deep learning in ecology. Methods in Ecology and Evolution, 10(10):1632–1644, 2019. 1

    work page 2019

Show all 42 references
  1. [9]

    Automated detection of wildlife using drones: Synthesis, opportunities and constraints

    Evangeline Corcoran, Megan Winsen, Ashlee Sudholz, and Grant Hamilton. Automated detection of wildlife using drones: Synthesis, opportunities and constraints. Methods in Ecology and Evolution, 12(6):1103–1114, 2021. 1

    2021

  2. [10]

    Cubaynes, Peter T

    Hannah C. Cubaynes, Peter T. Fretwell, Connor Bamford, Laura Gerrish, and Jennifer A. Jackson. Whales from space: Four mysticete species described using new vhr satellite im- agery. Marine Mammal Science, 35(2):466–491, 2019. 1

    2019

  3. [11]

    Alexandre Delplanque, Samuel Foucher, J ´erˆome Th´eau, Elsa Bussi`ere, C ´edric Vermeulen, and Philippe Lejeune. From crowd to herd counting: How to precisely detect and count african mammals using aerial imagery and deep learning? ISPRS Journal of Photogrammetry and Remote S...

    2023

  4. [12]

    Mac- donald, and Tiejun Wang

    Isla Duporge, Olga Isupova, Steven Reece, David W. Mac- donald, and Tiejun Wang. Using very-high-resolution satel- lite imagery and deep learning to detect and count African elephants in heterogeneous landscapes. Remote Sensing in Ecology and Conservation, 7(3):369–381, 2021. 1

    2021

  5. [13]

    BaboonLand Dataset: Tracking pri- mates in the wild and automating behaviour recognition from drone videos, 2024

    Isla Duporge, Maksim Kholiavchenko, Roi Harel, Scott Wolf, Dan Rubenstein, Meg Crofoot, Tanya Berger-Wolf, Stephen Lee, Julie Barreau, Jenna Kline, Michelle Ramirez, and Charles Stewart. BaboonLand Dataset: Tracking pri- mates in the wild and automating behaviour recognition f...

    2024

  6. [14]

    Jasper A. J. Eikelboom, Johan Wind, Eline van de Ven, Lek- ishon M. Kenana, Bradley Schroder, Henrik J. de Knegt, Frank van Langevelde, and Herbert H. T. Prins. Improving the precision and accuracy of animal population estimates with aerial image object detection. Methods in E...

    2019

  7. [15]

    Fretwell, Paul Scofield, and Richard A

    Peter T. Fretwell, Paul Scofield, and Richard A. Phillips. Us- ing super-high resolution satellite imagery to census threat- ened albatrosses. Ibis, 159(3):481–490, 2017. 1

    2017

  8. [16]

    Gray, Kevin C

    Patrick C. Gray, Kevin C. Bierlich, Sydney A. Mantell, Ari S. Friedlaender, Jeremy A. Goldbogen, and David W. John- ston. Drones and convolutional neural networks facilitate automated and accurate cetacean species identification and photogrammetry. Methods in Ecology and Evolu...

    2019

  9. [17]

    Gray, Abram B

    Patrick C. Gray, Abram B. Fleishman, David J. Klein, Matthew W. McKown, Vanessa S. B ´ezy, Kenneth J. Lohmann, and David W. Johnston. A convolutional neural network for detecting sea turtles in drone imagery. Methods in Ecology and Evolution, 10(3):345–355, 2019. 1

    2019

  10. [18]

    Drones and deep learning produce accurate and efficient monitoring of large- scale seabird colonies

    Madeline C Hayes, Patrick C Gray, Guillermo Harris, Wade C Sedgwick, Vivon D Crawford, Natalie Chazal, Sarah Crofts, and David W Johnston. Drones and deep learning produce accurate and efficient monitoring of large- scale seabird colonies. Ornithological Applications, 123(3): ...

    2021

  11. [19]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. CoRR, abs/1512.03385, 2015. 2

    2015 arXiv

  12. [20]

    Pytorch-Wildlife: A collab- orative deep learning framework for conservation, 2024

    Andres Hernandez, Zhongqi Miao, Luisa Vargas, Sara Beery, Rahul Dodhia, and Juan Lavista. Pytorch-Wildlife: A collab- orative deep learning framework for conservation, 2024. 5

    2024

  13. [21]

    Hodgson, Rowan Mott, Shane M

    Jarrod C. Hodgson, Rowan Mott, Shane M. Baylis, Trung T. Pham, Simon Wotherspoon, Adam D. Kilpatrick, Ramesh Raja Segaran, Ian Reid, Aleks Terauds, and Lian Pin Koh. Drones count wildlife more accurately and precisely than hu- mans. Methods in Ecology and Evolution , 9(5):1160–1167,

  14. [22]

    Application of Deep-Learning meth- ods to bird detection using unmanned aerial vehicle imagery

    Suk-Ju Hong, Yunhyeok Han, Sang-Yeon Kim, Ah-Yeong Lee, and Ghiseok Kim. Application of Deep-Learning meth- ods to bird detection using unmanned aerial vehicle imagery. Sensors (Basel), 19(7), 2019. 2

    2019

  15. [23]

    Thalassarche salvini

    BirdLife International. Thalassarche salvini. the IUCN red list of threatened species 2018: e.t22698388a132644161,

    2018

  16. [24]

    Eudyptes sclateri

    BirdLife International. Eudyptes sclateri. the IUCN red list of threatened species 2020: e.t22697789a131879000, 2020. Accessed on 10 March 2025. 5

    2020

  17. [25]

    Babu, Namrata Banerji, Elizabeth Campolongo, Matthew Thompson, Nina Van Tiel, Jackson Miliko, Ed- uardo Bessa, Majid Mirmehdi, Thomas Schmid, Tanya Berger-Wolf, Daniel I

    Maksim Kholiavchenko, Jenna Kline, Maksim Kukushkin, Otto Brookes, Sam Stevens, Isla Duporge, Alec Sheets, Reshma R. Babu, Namrata Banerji, Elizabeth Campolongo, Matthew Thompson, Nina Van Tiel, Jackson Miliko, Ed- uardo Bessa, Majid Mirmehdi, Thomas Schmid, Tanya Berger-Wolf,...

    2024

  18. [26]

    LaRue, Heather J

    Michelle A. LaRue, Heather J. Lynch, Phil O. B. Lyver, Kerry Barton, David G. Ainley, Annie Pollard, William R. Fraser, and Grant Ballard. A method for estimating colony sizes of Ad´elie penguins using remote sensing imagery. Po- lar Biology, 37(4):507–517, 2014. 1

    2014

  19. [27]

    Focal loss for dense object detection, 2018

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Doll´ar. Focal loss for dense object detection, 2018. 2

    2018

  20. [28]

    UA V equipped with infrared imaging for Cervi- dae monitoring: Improving detection accuracy by eliminat- ing background information interference

    Guangkai Ma, Wenjiao Li, Heng Bao, Nathan James Roberts, Yang Li, Weihua Zhang, Kun Yang, and Guang- shun Jiang. UA V equipped with infrared imaging for Cervi- dae monitoring: Improving detection accuracy by eliminat- ing background information interference. Ecological Infor- ...

    2024

  21. [29]

    An improved bird detection method using surveillance videos from Poyang Lake based on YOLOv8

    Jianchao Ma, Jiayuan Guo, Xiaolong Zheng, and Chaoyang Fang. An improved bird detection method using surveillance videos from Poyang Lake based on YOLOv8. Animals, 14 (23), 2024. 2

    2024

  22. [30]

    Sagar, and David R

    Thomas Mattern, Kalinka Rexer-Huber, Graham Parker, Jacinda Amey, Cara-Paige Green, Alan J.D Tennyson, Paul M. Sagar, and David R. Thompson. Erect-crested pen- guins on the Bounty Islands: population size and trends de- termined from ground counts and drone surveys. Notornis, ...

    2021

  23. [31]

    Mattern, David M

    Thomas Mattern, Klemens P ¨utz, Hannah L. Mattern, David M. Houston, Robin Long, Bianca C. Keys, Jeff W. White, Ursula Ellenberg, and Pablo Garcia-Borboroglu. Ac- curate abundance estimation of cliff-breeding Bounty Island 6 shags using drone-based 2D and 3D photogrammetry. Av...

    2023

  24. [32]

    Automated wildlife bird detection from drone footage using computer vision techniques

    Dimitrios Mpouziotas, Petros Karvelis, Ioannis Tsoulos, and Chrysostomos Stylios. Automated wildlife bird detection from drone footage using computer vision techniques. Ap- plied Sciences, 13(13), 2023. 2

    2023

  25. [33]

    Drone-based Salvin’s albatross population assessment: feasibility at the Bounty Islands

    Graham Parker and Kalinka Rexer-Huber. Drone-based Salvin’s albatross population assessment: feasibility at the Bounty Islands. Final report to Department of Conserva- tion, Marine Species and Threats. Parker Conservation, 126,

  26. [34]

    Phillips, R

    R.A. Phillips, R. Gales, G.B. Baker, M.C. Double, M. Favero, F. Quintana, M.L. Tasker, H. Weimerskirch, M. Uhart, and A. Wolfaardt. The conservation status and pri- orities for albatrosses and large petrels. Biological Conser- vation, 201:169–183, 2016. 2

    2016

  27. [35]

    Genetic programming with convolutional opera- tors for albatross nest detection from satellite imaging

    Mitchell Rogers, Igor Debski, Johannes Fischer, Peter Mc- Comb, Peter Frost, Bing Xue, Mengjie Zhang, and Patrice Delmas. Genetic programming with convolutional opera- tors for albatross nest detection from satellite imaging. In Advanced Concepts for Intelligent Vision Systems...

    2023

  28. [36]

    Benchmarking wild bird detection in complex forest scenes

    Qi Song, Yu Guan, Xi Guo, Xinhui Guo, Yufeng Chen, Hongfang Wang, Jianping Ge, Tianming Wang, and Lei Bao. Benchmarking wild bird detection in complex forest scenes. Ecological Informatics, 80:102466, 2024. 2

    2024

  29. [37]

    Salvin’s albatrosses at the Bounty Islands: at-sea dis- tribution

    David Thompson, Paul Sagar, Leigh Torres, and Matt Char- teris. Salvin’s albatrosses at the Bounty Islands: at-sea dis- tribution. Report prepared for New Zealand Department of Conservation, 2014. 2

    2014

  30. [38]

    Costelloe, Silvia Zuffi, Benjamin Risse, Alexander Mathis, Mackenzie W

    Devis Tuia, Benjamin Kellenberger, Sara Beery, Blair R. Costelloe, Silvia Zuffi, Benjamin Risse, Alexander Mathis, Mackenzie W. Mathis, Frank van Langevelde, Tilo Burghardt, Roland Kays, Holger Klinck, Martin Wikel- ski, Iain D. Couzin, Grant van Horn, Margaret C. Crofoot, Cha...

    2022

  31. [39]

    Weinstein, Sergio Marconi, Stephanie Bohlman, Alina Zare, and Ethan White

    Ben G. Weinstein, Sergio Marconi, Stephanie Bohlman, Alina Zare, and Ethan White. Individual tree-crown detec- tion in RGB imagery using semi-supervised deep learning neural networks. Remote Sensing, 11(11), 2019. 2, 4

    2019

  32. [40]

    Weinstein, Lindsey Garner, Vienna R

    Ben G. Weinstein, Lindsey Garner, Vienna R. Saccomanno, Ashley Steinkraus, Andrew Ortega, Kristen Brush, Glenda Yenni, Ann E. McKellar, Rowan Converse, Christopher D. Lippitt, Alex Wegmann, Nick D. Holmes, Alice J. Edney, Tom Hart, Mark J. Jessopp, Rohan H. Clarke, Dominik Mar...

    2022

  33. [41]

    Hughey, Jared A

    Zijing Wu, Ce Zhang, Xiaowei Gu, Isla Duporge, Lacey F. Hughey, Jared A. Stabach, Andrew K. Skidmore, J. Grant C. Hopcraft, Stephen J. Lee, Peter M. Atkinson, Douglas J. McCauley, Richard Lamprey, Shadrack Ngene, and Tiejun Wang. Deep learning enables satellite-based monitorin...

  34. [2018]

    Accessed on 10 March 2025. 2

    2025

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.