Pith. sign in

REVIEW 1 major objections 4 minor 65 references

Prompt-Tuning SAM: From Generalist to Specialist with only 2048 Parameters and 16 Training Images

T0 review · 1 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Adding 2,048 trainable prompt tokens to SAM's frozen mask decoder matches specialists that fine-tune millions of parameters.

desk verdict Useful parameter-efficient SAM adaptation, but the 'on par' claim for the 2,048-parameter variant is oversold; the image-encoder-tuned version is the real result. read the letter →

arxiv 2504.16739 v1 pith:WZ5AYVME submitted 2025-04-23 cs.CV

Tristan Piater , Björn Barz , Alexander Freytag This is my paper · ORCID
classification cs.CV
keywords prompttuningSegmentAnythingModelparameter-efficientfine-tuningbiomedicalimagesegmentationmicroscopyfew-shotlearningdomainshift
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 generalist segmentation model trained on natural images can be turned into a specialist for a specific biomedical imaging task without updating any of its pretrained weights. The authors propose PTSAM, which keeps the Segment Anything Model frozen and inserts a small number of learnable prompt vectors into the mask decoder's token stream, replacing the user prompts with a task-specific token. They report that 2,048 such parameters yield segmentation accuracy on par with established adaptation methods that train millions of parameters, across three microscopy datasets and one medical dataset. When additional prompts are inserted into every layer of the image encoder, accuracy improves further, and the method remains reliable with as few as 16 annotated training images. The potential payoff is that a single frozen foundation model could power many narrow-use-case segmentation systems at a fraction of the usual data and compute cost.

What carries the argument

The central object is the trainable prompt token: a real-valued vector inserted into the input sequence of a transformer layer. PTSAM inserts eight 256-dimensional prompts into the mask decoder's token stream (2,048 parameters) and, optionally, eight 768-dimensional prompts into each of the twelve layers of SAM's image encoder (73,728 parameters), following the deep visual prompt-tuning recipe. The prompts are learned while all of SAM's weights stay frozen, so they act as a compact steering signal that redirects the pretrained features toward the target segmentation task and domain. The mask decoder itself is a two-layer transformer that fuses image embeddings with a token stream; by changing only the token stream, PTSAM converts a promptable generalist into an automatic specialist.

What would settle it

Train PTSAM on a domain with a larger shift than the four tested, such as volumetric electron microscopy stacks or hyperspectral images, and compare against full fine-tuning; if more than eight mask-decoder prompts (or more than 73,728 image-encoder parameters) are needed to reach comparable Dice scores, or if PTSAM's accuracy falls significantly below the LoRA-based baselines, the claim that 2,048 parameters suffice for a specialist would be refuted.

Watch

Extended reading notes

Core claim

PTSAM removes SAM's prompt encoder and instead prepends eight learnable 256-dimensional vectors to the token stream consumed by SAM's two-layer mask-decoder transformer, keeping the no-prompt embedding and the three output tokens. Because the decoder can then generate a mask for every image without a point or box prompt, the model becomes an automatic task-specific segmenter. With only these 2,048 parameters trainable, PTSAM matches the Dice scores of AutoSAM, SAMed, and CellSeg1 on the four datasets while using roughly two thousand times fewer trainable parameters. When the same idea is extended to the image encoder, inserting eight learnable 768-dimensional prompts into each of the twelve ViT layers adds 73,728 parameters and improves Dice scores consistently, with the largest gains on the medical dataset where object boundaries are soft. Across all settings, PTSAM is the adaptation method least affected when training images are reduced from 64 to 16, which the authors attribute to the small parameter footprint preventing overfitting.

Load-bearing premise

The method assumes that SAM's frozen features, trained on natural images, are general enough that a handful of learnable vectors added to the token stream can steer them into a new domain; if a target domain is so far from natural images that this small steering capacity is insufficient, the 2,048-parameter claim would not hold.

Editorial extensions

If this is right

  • A frozen SAM becomes a deployable specialist for a new microscopy or medical task with 2,048 trainable parameters, so adapted models can be stored and shared as a few kilobytes instead of multi-million-parameter checkpoints.
  • Because the image encoder can also be prompt-tuned with 75,776 parameters, domain gaps such as soft-edged medical objects can be bridged with roughly a thousand times fewer parameters than full fine-tuning.
  • The robustness at 16 training images implies that new segmentation use cases with scarce annotations can be addressed without the overfitting seen in LoRA-based or fully fine-tuned adaptations.
  • Removing the need for per-image user prompts means the method runs fully automated in high-throughput pipelines, which is a prerequisite for many life-science applications.

Reading between the lines

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

  • A testable implication the paper leaves implicit is that the number of prompts needed scales with the domain distance from natural images; datasets such as volumetric electron microscopy or hyperspectral microscopy, which are further from SAM's training data, may require more than eight tokens per layer.
  • Another implicit consequence is that prompt tuning perturbs the frozen representation in a gentler way than low-rank weight updates, since PTSAM was the only method that consistently improved when the image encoder was adapted; if that is true for other foundation models, prompt tokens may be preferred over LoRA for cross-modal or cross-domain steering.
  • Because the method removes the prompt encoder entirely, it might also be combined with a lightweight prompt predictor to recover instance-level prompting when needed, while keeping the automatic mode for semantic tasks; the paper does not explore this hybrid.
  • The small parameter footprint also suggests that PTSAM could be applied per user or per server without storing full models, enabling personalized segmentation models at negligible storage cost.
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

1 major / 4 minor

Summary. The paper proposes PTSAM, a prompt-tuning method for adapting the Segment Anything Model (SAM) to domain-specific semantic segmentation by inserting learnable prompt tokens into SAM's mask decoder (2,048 trainable parameters) and optionally into every layer of the image encoder (75,776 total trainable parameters), while keeping all original SAM weights frozen. The method is evaluated on three microscopy datasets (EMParticles, LiveCell, OrganoidBasic) and one medical dataset (KvasirSEG) with training subsets of 16, 32, or 64 images, and compared against AutoSAM, CellSeg1, SAMed, and nnU-Net. The main claims are that mask-decoder-only PTSAM achieves accuracy on par with state-of-the-art adaptations while using roughly 2,000x fewer trainable parameters, and that additionally prompt-tuning the image encoder improves accuracy by up to 18–21% and remains robust in limited-data settings.

Significance. If the claims are properly supported, the contribution is practically valuable: it shows that prompt-tuning can remove SAM's prompt-dependence and adapt it to non-natural domains with very few trainable parameters, which is attractive for microscopy and medical scenarios with scarce annotations. The paper has clear strengths: explicit parameter counts (Table 1), evaluation on four datasets across three training-set sizes, a held-out ablation for the prompt counts, and a simple architecture that is easy to reproduce. The central caveat is that the headline parity claim for the 2,048-parameter variant is not yet established by the reported statistics; the qualitative and quantitative evidence for the image-encoder-prompted variant is stronger.

major comments (1)
  1. [§4.2.1 (Table 2)] The conclusion that PTSAM with a frozen image encoder is 'on par' with state-of-the-art methods is not supported by the reported statistics. In 10 of the 12 frozen-IE conditions in Table 2, PTSAM's mean Dice is below AutoSAM's, with differences such as 87.7 vs 91.8 (EMParticles-16), 86.4 vs 89.4 (LiveCell-64), and 89.4 vs 90.9 (OrganoidBasic-16) that are often larger than the per-condition standard deviations. With only three random splits, absence of statistical significance is expected under low power and does not justify an equivalence claim; the consistent direction of the deficit is evidence of a systematic gap. Please either add an equivalence test with a pre-specified margin or reformulate the claim as 'slightly lower accuracy at roughly 2,000x fewer trainable parameters.'
minor comments (4)
  1. [Abstract and §1] The Abstract claims 'up to 18%' improvement over state-of-the-art results, while contribution 2 in §1 claims 'up to 21%'. Please reconcile the two numbers and specify whether the improvement is expressed in absolute percentage points or relative Dice gain, and against which baseline it is computed.
  2. [§4.1.3 and §5] The paper reports tuning learning rates per method and selecting n_md and n_ie on a held-out dataset, yet §5 concludes that PTSAM requires 'no hyperparameters to tune'. This should be qualified to avoid overstating the method's practicality.
  3. [Table 2 and §4.1.2] Please consider reporting the total number of images per dataset or providing this information in the dataset descriptions, since the 16/32/64 training subsets otherwise lack context.
  4. [§4.2.4] Please add a space in 'by4x' in the first paragraph of §4.2.4.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: PTSAM is an empirical prompt-tuning adaptation evaluated against external baselines on held-out splits.

full rationale

The derivation chain here is an experimental method, not a mathematical derivation. PTSAM's trainable prompts (p_md and p_ie) are inserted into a frozen SAM and optimized with a standard loss (0.2 CE + 0.8 Dice) on training splits; the reported Dice scores are measured on held-out test splits (20% of each dataset) and compared against external methods such as AutoSAM, CellSeg1, SAMed, and nnU-Net. There is no equation in the paper that defines the method's success metric in terms of its own inputs, and no parameter is fitted to the test data: hyperparameters, including nmd = nie = 8 and the learning rates, were chosen on a held-out dataset and kept fixed. The prompt vectors are trained, not derived from the evaluation results, so the central efficiency claim (2,048 parameters) is an empirical finding rather than a tautology. The skeptical concern that 'on par' is asserted from non-significant differences despite systematically lower scores is an issue of statistical interpretation and empirical support, not circular reasoning; under the stated rules, underpowered statistics or lack of equivalence testing belongs in correctness risk, not in the circularity score. No load-bearing self-citations, imported uniqueness theorems, or ansatz-smuggling citations were found; visual prompt tuning [30] and SAM [31] are external, independent foundations, and the paper is self-contained against external benchmarks.

Assumptions & free parameters 3 free parameters · 1 assumptions · 0 invented entities

The paper introduces no new physical or mathematical entities. The free parameters are the number of prompts and learning rates, tuned on a held-out dataset. The only domain assumption is that SAM's frozen features transfer to the target domains, which the experiments are designed to test.

free parameters (3)
  • n_prompts (nmd = nie = 8) = 8
    Chosen from ablation experiments on a held-out dataset (Sec. 4.3); the central '2,048 parameters' claim uses nmd = 8.
  • learning rate (decoder-only) = 0.05
    Per-method learning rate tuned on a held-out dataset (Sec. 4.1.3).
  • learning rate (with image encoder) = 0.01
    Per-method learning rate tuned on a held-out dataset (Sec. 4.1.3).
assumptions (1)
  • domain assumption SAM's pretrained ViT-b weights, trained on natural images, provide a useful prior for microscopy and medical images.
    The entire method relies on the frozen SAM features being transferable; if the domain gap were too large, prompt-only adaptation would not work. This assumption is implicit in Section 3.2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Prompt-Tuning SAM: From Generalist to Specialist with only 2048 Parameters and 16 Training Images." pith.science (2026). https://pith.science/paper/WZ5AYVME

@misc{pith2026250416739,
  author       = {Pith},
  title        = {Pith review of: Prompt-Tuning SAM: From Generalist to Specialist with only 2048 Parameters and 16 Training Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WZ5AYVME}},
  note         = {Machine review of arXiv:2504.16739}
}
read the original abstract

The Segment Anything Model (SAM) is widely used for segmenting a diverse range of objects in natural images from simple user prompts like points or bounding boxes. However, SAM's performance decreases substantially when applied to non-natural domains like microscopic imaging. Furthermore, due to SAM's interactive design, it requires a precise prompt for each image and object, which is unfeasible in many automated biomedical applications. Previous solutions adapt SAM by training millions of parameters via fine-tuning large parts of the model or of adapter layers. In contrast, we show that as little as 2,048 additional parameters are sufficient for turning SAM into a use-case specialist for a certain downstream task. Our novel PTSAM (prompt-tuned SAM) method uses prompt-tuning, a parameter-efficient fine-tuning technique, to adapt SAM for a specific task. We validate the performance of our approach on multiple microscopic and one medical dataset. Our results show that prompt-tuning only SAM's mask decoder already leads to a performance on-par with state-of-the-art techniques while requiring roughly 2,000x less trainable parameters. For addressing domain gaps, we find that additionally prompt-tuning SAM's image encoder is beneficial, further improving segmentation accuracy by up to 18% over state-of-the-art results. Since PTSAM can be reliably trained with as little as 16 annotated images, we find it particularly helpful for applications with limited training data and domain shifts.

Figures

Figures reproduced from arXiv: 2504.16739 by the authors.

Figure 2
Figure 2. Overview of our PTSAM approach. We remove the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Comparing improvements of different methods when [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 5
Figure 5. Ablation: investigating the effect of different numbers of [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figures from the paper (1 more)
Figure 6
Figure 6. Figure 6: Qualitative segmentation results on three microscopy and one medical dataset (top to bottom). On each dataset, one example [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 49 canonical work pages

  1. [1]

    GPT-4 technical report

    Josh Achiam et al. GPT-4 technical report. arXiv preprint arXiv:2303.08774, 2023. 1, 2

    arXiv 2023

  2. [2]

    Image analysis for mri based brain tumor detection

    Sidra Akbar, Shahzad Nasim, Sarwar Wasi, and Syed Muhammad Ussama Zafar. Image analysis for mri based brain tumor detection. 2019 4th International Conference on Emerging Trends in Engineering, Sciences and Technology (ICEEST), pages 1–5, 2019. 1

    work page 2019

  3. [3]

    Al-antari, Jamil Hussain, Nag- wan Abdel Samee, Maali Alabdulhafith, and Akio Gofuku

    Asaad Anaam, Mugahed A. Al-antari, Jamil Hussain, Nag- wan Abdel Samee, Maali Alabdulhafith, and Akio Gofuku. Deep active learning for automatic mitotic cell detection on hep-2 specimen medical images. Diagnostics, 13, 2023. 1

    work page 2023

  4. [4]

    Segment anything for microscopy

    Anwai Archit, Luca Freckmann, Sushmita Nair, Nabeel Khalid, Paul Hilt, Vikas Rajashekar, Marei Freitag, Carolin Teuber, Genevieve Buckley, Sebastian von Haaren, et al. Segment anything for microscopy. Nature Methods, pages 1–13, 2025. 2, 3

    work page 2025

  5. [5]

    Strong baselines for parameter-efficient few-shot fine-tuning

    Samyadeep Basu, Shell Xu Hu, Daniela Massiceti, and So- heil Feizi. Strong baselines for parameter-efficient few-shot fine-tuning. In Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI 2024, Thirty-Sixth Conference on Innova- tive Applications of Artificial Intelligence, IAAI 2024, Four- teenth Symposium on Educational Advances in Artificial I...

    work page 2024

  6. [6]

    Hudson, Ehsan Adeli, et al

    Rishi Bommasani, Drew A. Hudson, Ehsan Adeli, et al. On the opportunities and risks of foundation models. ArXiv preprint, abs/2108.07258, 2021. 1, 2

    arXiv 2021

  7. [7]

    Brown, Benjamin Mann, Nick Ryder, et al

    Tom B. Brown, Benjamin Mann, Nick Ryder, et al. Lan- guage models are few-shot learners. In Advances in Neu- ral Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual, 2020. 1, 2

    work page 2020

  8. [8]

    Guttag, and Adrian V

    Victor Ion Butoi, Jose Javier Gonzalez Ortiz, Tianyu Ma, Mert Rory Sabuncu, John V . Guttag, and Adrian V . Dalca. Universeg: Universal medical image segmentation. 2023 IEEE/CVF International Conference on Computer Vision (ICCV), pages 21381–21394, 2023. 2

    work page 2023

Show all 65 references
  1. [9]

    Tinytl: Reduce activations, not trainable parameters for efficient on- device learning

    Han Cai, Chuang Gan, Ligeng Zhu, and Song Han. Tinytl: Reduce activations, not trainable parameters for efficient on- device learning. ArXiv preprint, abs/2007.11622, 2020. 2

    2007 arXiv

  2. [10]

    Segment anything model (sam) for digital pathology: Assess zero- shot segmentation on whole slide imaging

    Ruining Deng, Can Cui, Quan Liu, Tianyuan Yao, Lu- cas W Remedios, Shunxing Bao, Bennett A Landman, Lee E Wheless, Lori A Coburn, Keith T Wilson, et al. Segment anything model (sam) for digital pathology: Assess zero- shot segmentation on whole slide imaging. ArXiv preprint, a...

    2023 arXiv

  3. [11]

    Qlora: Efficient finetuning of quantized llms

    Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. Qlora: Efficient finetuning of quantized llms. In Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Sys- tems 2023, NeurIPS 2023, New Orleans, LA, USA, De...

    2023

  4. [12]

    Decaf: A deep convolutional activation feature for generic visual recogni- tion

    Jeff Donahue, Yangqing Jia, Oriol Vinyals, Judy Hoffman, Ning Zhang, Eric Tzeng, and Trevor Darrell. Decaf: A deep convolutional activation feature for generic visual recogni- tion. In Proceedings of the 31th International Conference on Machine Learning, ICML 2014, Beijing, Ch...

    2014

  5. [13]

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

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition ...

    2021

  6. [14]

    Jackson, Nabeel Khalid, Nicola J Bevan, Timothy Dale, Andreas R

    Christoffer Edlund, Timothy R. Jackson, Nabeel Khalid, Nicola J Bevan, Timothy Dale, Andreas R. Dengel, Sheraz Ahmed, Johan Trygg, and Rickard Sj ¨ogren. Livecell—a large-scale dataset for label-free live cell segmentation. Na- ture Methods, 18:1038 – 1045, 2021. 5

    2021

  7. [15]

    Parameter-efficient fine-tuning for large models: A comprehensive survey

    Zeyu Han, Chao Gao, Jinyang Liu, Jeff Zhang, and Sai Qian Zhang. Parameter-efficient fine-tuning for large models: A comprehensive survey. ArXiv preprint, abs/2403.14608,

    arXiv

  8. [16]

    Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images

    Ali Hatamizadeh, Vishwesh Nath, Yucheng Tang, Dong Yang, Holger R Roth, and Daguang Xu. Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images. In International MICCAI brainlesion workshop, pages 272–284. Springer, 2021. 2

    2021

  9. [17]

    UNETR: Transformers for 3d medical im- age segmentation, 2021

    Ali Hatamizadeh, Yucheng Tang, Vishwesh Nath, Dong Yang, Andriy Myronenko, Bennett Landman, Holger Roth, and Daguang Xu. UNETR: Transformers for 3d medical im- age segmentation, 2021. 3

    2021

  10. [18]

    Unetr: Transformers for 3d med- ical image segmentation

    Ali Hatamizadeh, Yucheng Tang, Vishwesh Nath, Dong Yang, Andriy Myronenko, Bennett Landman, Holger R Roth, and Daguang Xu. Unetr: Transformers for 3d med- ical image segmentation. In Proceedings of the IEEE/CVF winter conference on applications of computer vision, pages 574–58...

    2022

  11. [19]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 2

    2016

  12. [20]

    Girshick

    Kaiming He, Georgia Gkioxari, Piotr Doll ´ar, and Ross B. Girshick. Mask R-CNN. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22- 29, 2017, pages 2980–2988. IEEE Computer Society, 2017. 2

    2017

  13. [21]

    Girshick

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Pi- otr Doll ´ar, and Ross B. Girshick. Masked autoencoders are scalable vision learners. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022 , pages 15979–15988....

    2022

  14. [22]

    Parameter-efficient transfer learning for NLP

    Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin de Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. Parameter-efficient transfer learning for NLP. In Proceedings of the 36th Interna- tional Conference on Machine Learning, ICML 2019, ...

    2019

  15. [23]

    Universal language model fine-tuning for text classification

    Jeremy Howard and Sebastian Ruder. Universal language model fine-tuning for text classification. In Proceedings of the 56th Annual Meeting of the Association for Computa- tional Linguistics (Volume 1: Long Papers), pages 328–339, Melbourne, Australia, 2018. Association for Com...

    2018

  16. [24]

    Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

    Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models. In The Tenth International Conference on Learning Represen- tations, ICLR 2022, Virtual Event, April 25-29, 2022....

    2022

  17. [25]

    How to efficiently adapt large segmentation model(sam) to medical images,

    Xinrong Hu, Xiaowei Xu, and Yiyu Shi. How to efficiently adapt large segmentation model(sam) to medical images,

  18. [26]

    Jaeger, Simon A

    Fabian Isensee, Jens Petersen, Andr ´e Klein, David Zim- merer, P. Jaeger, Simon A. A. Kohl, J. Wasserthal, Gregor Koehler, T. Norajitra, Sebastian J. Wirkert, and Klaus Her- mann Maier-Hein. nnu-net: Self-adapting framework for u-net-based medical image segmentation. ArXiv pr...

    2018 arXiv

  19. [27]

    A foundation model for cell segmentation, 2023

    Uriah Israel, Markus Marks, Rohit Dilip, Qilin Li, Morgan Schwartz, Elora Pradhan, Edward Pao, Shenyi Li, Alexan- der Pearson-Goulart, Pietro Perona, Georgia Gkioxari, Ross Barnowski, Yisong Yue, and David Van Valen. A foundation model for cell segmentation, 2023. 2, 3, 5

    2023

  20. [28]

    Riegler, P

    Debesh Jha, Pia Helen Smedsrud, M. Riegler, P. Halvorsen, Thomas de Lange, Dag Johansen, and Haavard D. Johansen. Kvasir-seg: A segmented polyp dataset. In Conference on Multimedia Modeling, 2019. 5

    2019

  21. [29]

    Segment anything is not always perfect: An investi- gation of sam on different real-world applications, 2024

    Wei Ji, Jingjing Li, Qi Bi, Tingwei Liu, Wenbo Li, and Li Cheng. Segment anything is not always perfect: An investi- gation of sam on different real-world applications, 2024. 2

    2024

  22. [30]

    Belongie, Bharath Hariharan, and Ser Nam Lim

    Menglin Jia, Luming Tang, Bor-Chun Chen, Claire Cardie, Serge J. Belongie, Bharath Hariharan, and Ser Nam Lim. Vi- sual prompt tuning. ArXiv preprint, abs/2203.12119, 2022. 2, 3, 4

    2022 arXiv

  23. [31]

    Berg, Wan-Yen Lo, Piotr Doll ´ar, and Ross B

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chlo´e Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C. Berg, Wan-Yen Lo, Piotr Doll ´ar, and Ross B. Girshick. Segment anything. In IEEE/CVF Inter- national Conference on Computer Vision, ICCV 202...

    2023

  24. [32]

    Lee, Jacob S

    Michael Y . Lee, Jacob S. Bedia, Salil S. Bhate, Graham L. Barlow, Darci Phillips, Wendy J. Fantl, Garry P. Nolan, and Christian M. Sch¨urch. Cellseg: a robust, pre-trained nucleus segmentation and pixel quantification software for highly multiplexed fluorescence images. BMC B...

  25. [33]

    van der Ent, Jeffrey M

    Juliet Lefferts, Suzanne Kroes, Matthew B Smith, Paul J Niem¨oller, Natascha D A Nieuwenhuijze, Heleen N Sonn- eveld van Kooten, Cornelis K. van der Ent, Jeffrey M. Beek- man, and Sam F B van Beuningen. Orgasegment: deep- learning based organoid segmentation to quantify cftr d...

    2024

  26. [34]

    The power of scale for parameter-efficient prompt tuning

    Brian Lester, Rami Al-Rfou, and Noah Constant. The power of scale for parameter-efficient prompt tuning. In Proceed- ings of the 2021 Conference on Empirical Methods in Nat- ural Language Processing , pages 3045–3059, Online and Punta Cana, Dominican Republic, 2021. Associatio...

    2021

  27. [35]

    Adapting segment anything models to medical imaging via fine-tuning without domain pretraining

    Kevin Li and Pranav Rajpurkar. Adapting segment anything models to medical imaging via fine-tuning without domain pretraining. In AAAI 2024 Spring Symposium on Clinical Foundation Models, 2024. 5

    2024

  28. [36]

    Prefix-tuning: Optimiz- ing continuous prompts for generation

    Xiang Lisa Li and Percy Liang. Prefix-tuning: Optimiz- ing continuous prompts for generation. In Proceedings of the 59th Annual Meeting of the Association for Computa- tional Linguistics and the 11th International Joint Confer- ence on Natural Language Processing (Volume 1: Lo...

    2021

  29. [37]

    Exploring plain vision transformer backbones for object de- tection

    Yanghao Li, Hanzi Mao, Ross Girshick, and Kaiming He. Exploring plain vision transformer backbones for object de- tection. In European conference on computer vision , pages 280–296. Springer, 2022. 4

    2022

  30. [38]

    Ex- ploring versatile generative language model via parameter- efficient transfer learning

    Zhaojiang Lin, Andrea Madotto, and Pascale Fung. Ex- ploring versatile generative language model via parameter- efficient transfer learning. In Findings of the Association for Computational Linguistics: EMNLP 2020 , pages 441–459, Online, 2020. Association for Computational Li...

    2020

  31. [39]

    Fully convolutional networks for semantic segmentation

    Jonathan Long, Evan Shelhamer, and Trevor Darrell. Fully convolutional networks for semantic segmentation. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2015, Boston, MA, USA, June 7-12, 2015 , pages 3431–3440. IEEE Computer Society, 2015. 2

    2015

  32. [40]

    SGDR: stochastic gradient descent with warm restarts

    Ilya Loshchilov and Frank Hutter. SGDR: stochastic gradient descent with warm restarts. In 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings. OpenRe- view.net, 2017. 5

    2017

  33. [41]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net, 2019. 5

    2019

  34. [42]

    Segment anything in medical images, 2023

    Jun Ma, Yuting He, Feifei Li, Lin Han, Chenyu You, and Bo Wang. Segment anything in medical images, 2023. 5, 6

    2023

  35. [43]

    Segment any cell: A sam-based auto- prompting fine-tuning framework for nuclei segmentation

    Saiyang Na, Yuzhi Guo, Feng Jiang, Hehuan Ma, and Junzhou Huang. Segment any cell: A sam-based auto- prompting fine-tuning framework for nuclei segmentation. ArXiv preprint, abs/2401.13220, 2024. 2, 3

    2024 arXiv

  36. [44]

    Retfound: Training robust medical imaging-based diagnostic models via self-supervised learning

    Laurentino Biccas Neto and Caio Regattieri. Retfound: Training robust medical imaging-based diagnostic models via self-supervised learning. Arquivos Brasileiros de Oftal- mologia, 2023. 2

    2023

  37. [45]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision. In Proceedings of th...

    2021

  38. [46]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. ArXiv preprint, abs/1505.04597, 2015. 2, 3

    2015 arXiv

  39. [47]

    AdapterDrop: On the efficiency of adapters in transformers

    Andreas R ¨uckl´e, Gregor Geigle, Max Glockner, Tilman Beck, Jonas Pfeiffer, Nils Reimers, and Iryna Gurevych. AdapterDrop: On the efficiency of adapters in transformers. In Proceedings of the 2021 Conference on Empirical Meth- ods in Natural Language Processing, pages 7930–79...

    2021

  40. [48]

    A large and diverse brain organoid dataset of 1,400 cross- laboratory images of 64 trackable brain organoids

    Julian Schr ¨oter, Luca Deininger, Blaz Lupse, Petra Richter, Steffen Syrbe, Ralf Mikut, and Sabine Jung-Klawitter. A large and diverse brain organoid dataset of 1,400 cross- laboratory images of 64 trackable brain organoids. Scientific Data, 11, 2024. 1

    2024

  41. [49]

    Cell segmentation and dry mass char- acterization based on slightly off-axis holographic imaging

    Qian Shen, Zhuoshi Li, Yuanyuan Chen, Haojie Gu, Qian Chen, and Chao Zuo. Cell segmentation and dry mass char- acterization based on slightly off-axis holographic imaging. In International Conference on Optical and Photonic Engi- neering, 2023. 1

    2023

  42. [50]

    A review paper on microscopic modalities used in biomedical applications

    Neeru Singla, Ajay Kumar, Himanshu Rikhari, and Vishal Srivastava. A review paper on microscopic modalities used in biomedical applications. In European Journal of Molecu- lar & Clinical Medicine, 2020. 1

    2020

  43. [51]

    Three things everyone should know about vision transformers

    Hugo Touvron, Matthieu Cord, Alaaeldin El-Nouby, Jakob Verbeek, and Herv ´e J ´egou. Three things everyone should know about vision transformers. In European Conference on Computer Vision, pages 497–515. Springer, 2022. 2

    2022

  44. [52]

    Go- ing beyond u-net: Assessing vision transformers for se- mantic segmentation in microscopy image analysis

    Illia Tsiporenko, Pavel Chizhov, and Dmytro Fishman. Go- ing beyond u-net: Assessing vision transformers for se- mantic segmentation in microscopy image analysis. ArXiv preprint, abs/2409.16940, 2024. 2

    2024 arXiv

  45. [53]

    Sam F. B. van Beuningen, Juliet Lefferts, and University Medical Center Utrecht. Organoid basic, 2023. 5

    2023

  46. [54]

    VandeLoo, Nathan J

    Alexandra D. VandeLoo, Nathan J. Malta, Emilio Aponte, Caitlin van Zyl, Danfei Xu, and Craig R. Forest. Samcell: Generalized label-free biological cell segmentation with seg- ment anything. bioRxiv, 2025. 2, 3, 4, 5

    2025

  47. [55]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need. In Advances in Neu- ral Information Processing Systems 30: Annual Conference on Neural Information Processing Systems 2017...

    2017

  48. [56]

    Anchor DETR: query design for transformer-based detec- tor

    Yingming Wang, Xiangyu Zhang, Tong Yang, and Jian Sun. Anchor DETR: query design for transformer-based detec- tor. In Thirty-Sixth AAAI Conference on Artificial Intelli- gence, AAAI 2022, Thirty-Fourth Conference on Innovative Applications of Artificial Intelligence, IAAI 2022...

    2022

  49. [57]

    Self-prompting large vision models for few-shot medical image segmenta- tion, 2023

    Qi Wu, Yuyao Zhang, and Marawan Elbatel. Self-prompting large vision models for few-shot medical image segmenta- tion, 2023. 3

    2023

  50. [58]

    Parameter-efficient fine-tuning for pre-trained vision models: A survey

    Yi Xin, Siqi Luo, Haodi Zhou, Junlong Du, Xiaohong Liu, Yue Fan, Qing Li, and Yuntao Du. Parameter-efficient fine-tuning for pre-trained vision models: A survey. ArXiv preprint, abs/2402.02242, 2024. 2

    2024

  51. [59]

    Joe Qin, Xiaohui Tao, and Fu Lee Wang

    Lingling Xu, Haoran Xie, S. Joe Qin, Xiaohui Tao, and Fu Lee Wang. Parameter-efficient fine-tuning methods for pretrained language models: A critical review and assess- ment. ArXiv preprint, abs/2312.12148, 2023. 2

    2023 arXiv

  52. [60]

    Depth anything: Unleashing the power of large-scale unlabeled data

    Lihe Yang, Bingyi Kang, Zilong Huang, Xiaogang Xu, Jiashi Feng, and Hengshuang Zhao. Depth anything: Unleashing the power of large-scale unlabeled data. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10371–10381, 2024. 2

    2024

  53. [61]

    Batuhan Yildirim and Jacqueline M. Cole. Bayesian particle instance segmentation for electron microscopy image quan- tification. Journal of Chemical Information and Modeling , 61(3):1136–1149, 2021. PMID: 33682402. 5

    2021

  54. [62]

    Customized segment any- thing model for medical image segmentation, 2023

    Kaidong Zhang and Dong Liu. Customized segment any- thing model for medical image segmentation, 2023. 3, 4, 5

    2023

  55. [63]

    Sixiao Zheng, Jiachen Lu, Hengshuang Zhao, Xiatian Zhu, Zekun Luo, Yabiao Wang, Yanwei Fu, Jianfeng Feng, Tao Xiang, Philip H. S. Torr, and Li Zhang. Rethinking seman- tic segmentation from a sequence-to-sequence perspective with transformers. In IEEE Conference on Computer Vi...

    2021

  56. [64]

    Cellseg1: Robust cell segmentation with one training image, 2024

    Peilin Zhou, Bo Du, and Yongchao Xu. Cellseg1: Robust cell segmentation with one training image, 2024. 3, 4, 5

    2024

  57. [2021]

    OpenReview.net, 2021. 2, 3

    2021

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.