Pith. sign in

REVIEW 4 major objections 8 minor 35 references

Phase-fraction guided denoising diffusion model for augmenting multiphase steel microstructure segmentation via micrograph image-mask pair synthesis

T0 review · 4 major / 8 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 PF-DiffSeg jointly synthesizes steel micrograph image-mask pairs conditioned on phase fractions, and augmenting segmentation training with those pairs lifts rare-phase IoU by more than ten points.

desk verdict A useful domain adaptation of joint image-mask diffusion with plausible but possibly inflated segmentation gains due to test-set selection of the augmentation ratio and missing uncertainty. read the letter →

arxiv 2508.00896 v1 pith:PSXOZZCW submitted 2025-07-28 cs.CV cond-mat.mtrl-scieess.IV

Hoang Hai Nam Nguyen , Minh Tien Tran , Hoheok Kim , Ho Won Lee This is my paper · ORCID
classification cs.CVcond-mat.mtrl-scieess.IV
keywords generativedataaugmentationdenoisingdiffusionprobabilisticmodelmicrostructuresegmentationmultiphasesteelphase-fractionconditioningsemanticrare-phaseoversamplingimage-maskpairsynthesis
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 is trying to establish that synthetic training data generated by a single diffusion model can replace hand-designed augmentation for segmenting multiphase steel micrographs. The key idea is to condition the generative trajectory on a global phase-fraction vector, so the model produces image-mask pairs with specified amounts of matrix, austenite, martensite/austenite, and defect, and to oversample rare phases during conditioning. If this works, it matters because pixel-accurate annotation of rare phases is the bottleneck in metallographic image analysis; a scalable generator that yields aligned pairs could cut labeling effort and support automated quality control.

What carries the argument

The load-bearing object is the concatenated image-mask tensor $\tilde{x}_0 = [x_0, y_0] \in \mathbb{R}^{128\times128\times4}$, treated as a single diffusion sample, together with the conditioning vector $c = [f_M, f_A, f_{MA}, f_D]$, the global phase fractions of matrix, austenite, martensite/austenite, and defect. A six-stage U-Net denoiser receives $c$ through a linear projection injected into every ResNet block, so global composition constraints act throughout the denoising trajectory. Training uses a hybrid MSE plus L1 noise-prediction loss, sampling uses 50-step DDIM, and synthetic condition vectors are drawn by jittering real fractions with 70% uniform, 20% MA-boosted, and 10% defect-boosted components. A separate iterative super-resolution model upsamples the generated 128x128 pairs to 256x256 for segmentation training, enforcing sharp phase boundaries while preserving the conditioned phase fractions.

What would settle it

Recompute the segmentation comparison using leave-one-micrograph-out cross-validation across all 42 MetalDAM micrographs; if the average MIoU gain over basic augmentation falls below the reported 3–5 points or reverses on the rare classes, the central claim is not robust.

Watch

Extended reading notes

Core claim

The central claim is that a one-stage denoising diffusion model can generate paired SEM micrographs and pixel-level phase masks in a single reverse-diffusion pass, conditioned only on a four-component phase-fraction vector, and that augmenting segmentation training with these synthetic pairs materially improves accuracy, especially for rare martensite/austenite islands and defects. On the MetalDAM benchmark, the approach reports MIoU gains of roughly 3–5 points over basic augmentation and 1.5–3.5 points over a two-stage mask-guided diffusion baseline, with per-class IoU for MA rising above 60 and Defect above 80 when minority-phase oversampling is used. The same joint formulation also removes the need for a separate mask-to-image translation stage, cutting end-to-end generation time by about 17 percent.

Load-bearing premise

The reported gains rest on the assumption that the five held-out micrographs fairly represent the MetalDAM distribution, so the measured 3–5 point MIoU improvement will carry over to other steel microstructures and imaging conditions.

Editorial extensions

If this is right

  • Across U-Net, U-Net++, LinkNet, and MA-Net, training on real plus PF-DiffSeg synthetic pairs raises MIoU and pixel accuracy relative to basic augmentation.
  • The largest per-class gains appear in the rare classes: MA IoU rises from about 41% to above 60% with boosting, and Defect IoU from about 50% to above 80%.
  • Synthetic-data scaling shows gains emerging at roughly four times the real data volume, peaking near eight times, after which performance plateaus from domain overfitting.
  • Joint one-stage generation avoids the spatial drift of mask-then-image pipelines and saves about 17% of end-to-end generation time per image-mask pair.
  • Phase-fraction conditioning lets the user rebalance the training distribution toward under-represented microconstituents without handcrafted oversampling of images.

Reading between the lines

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

  • Beyond the paper's experiments, the same phase-fraction conditioning should transfer to other alloy systems and imaging modalities such as optical microscopy, EBSD, or TEM, because the conditioning signal is a physically interpretable composition descriptor rather than an image-specific feature.
  • An editorial extension: because the conditioning vector encodes only global composition, adding shape statistics such as grain size, aspect ratio, or spatial phase adjacency would likely give finer control over generated morphology and could push segmentation gains further.
  • The observed plateau and overfitting at high synthetic volumes suggests a testable improvement: filtering generated pairs by fidelity or applying a small domain-adaptation step before augmentation could extend the useful scaling range beyond the reported 8x point.
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

4 major / 8 minor

Summary. PF-DiffSeg is a phase-fraction-conditioned denoising diffusion model that generates paired SEM micrograph images and segmentation masks in a single reverse-diffusion pass. The authors train the model on MetalDAM steel micrographs, sample synthetic conditioning vectors by perturbing real phase fractions and oversampling rare MA/defect classes, upscale the generated 128×128 pairs to 256×256 with a separate super-resolution diffusion model, and use the synthetic pairs to augment training of four segmentation architectures. They report that PF-DiffSeg augmentation improves MIoU by 3–5 points over basic augmentation and 1.5–3.5 points over the two-stage SegGuidedDiff baseline, raises rare-phase IoU by more than 10 points, and reduces end-to-end generation time by about 17% compared with SegGuidedDiff, and they compare favorably with a DCGAN+Pix2PixHD baseline reported in prior work.

Significance. The proposed conditioning mechanism is physically interpretable, and the one-stage joint image-mask synthesis is a sensible way to avoid the semantic drift of two-stage pipelines. If the reported gains replicate under a properly controlled evaluation, the method would be a practical tool for segmentation of imbalanced metallographic datasets. The paper shows consistent MIoU improvements across four architectures, provides a public code link, and gives a clear description of the generative framework. However, the current evaluation protocol does not yet support the headline numbers: the augmentation ratio is selected on the held-out test set, no uncertainty is reported in Table 3, and the GAN comparison is not controlled. These issues are fixable and do not invalidate the approach, but they require a major revision before the claims can be accepted.

major comments (4)
  1. [Section 2.3, Fig. 9] The reported MIoU gains are selected over augmentation ratios on the held-out test set. Because the text says "only highest metrics were reported" for ratios 1:1–1:10 and Fig. 9 plots MIoU versus synthetic volume on the same five-micrograph test set, the Table 3 numbers are effectively test-set-selected. Please choose the real-to-synthetic ratio on a validation split (e.g., one or two training micrographs held out) and then report test-set metrics for that fixed ratio; alternatively, report all ratios and show that the gains are insensitive to the choice.
  2. [Section 3.2, Table 3] The paper states that all experiments were run five times with different random seeds, but Table 3 reports only point estimates. The headline gains of 3–5% MIoU and more than 10% rare-phase IoU may be smaller than run-to-run and image-to-image variability, especially with only five test micrographs. Please report mean ± standard deviation over the five runs, or per-image IoU values, and indicate whether the differences are statistically meaningful (e.g., a paired test per image).
  3. [Table 4] The DCGAN+Pix2PixHD comparison quotes per-class IoU values from [5] for validation images A and B rather than evaluating the baseline in the same experimental pipeline. Because the segmentation models, training data, and resolution differ, this comparison is uncontrolled and cannot support the claim that PF-DiffSeg outperforms the GAN baseline. Either implement DCGAN+Pix2PixHD under the same training protocol or clearly frame Table 4 as "as reported in [5]" and avoid comparative claims based on it.
  4. [Section 3.1, Fig. 4] The phase-fraction adherence check measures agreement between generated mask fractions and the conditioning vector that the model is explicitly trained to satisfy. This is an internal consistency check, not independent evidence of compositional validity or physical plausibility. Please state this limitation in the text; the downstream segmentation improvement is the appropriate validation, but it is currently entangled with the test-set selection issue raised above.
minor comments (8)
  1. [Section 2.2.1, Eq. (2)] The conditional distribution is written as p_θ(x̃_{t-1} | x̃_{t-1}, c); it should condition on x̃_t. Please correct this typo.
  2. [Section 2.2.1 and Section 2.2.3] Equation (5) is used twice, once for the DDIM update and once for the super-resolution generation; renumber the equations to avoid ambiguity.
  3. [Section 2.2.1, step 2] The phrase "exactly the desired phase composition" overstates the control shown in Fig. 4, where the reported MAE for Austenite is 1.68%; please soften this wording.
  4. [Table 1] Precipitate is listed as a class in Table 1, but Section 2.1 says it is ignored; either remove it from the table or note that it is excluded from all model inputs and metrics.
  5. [Fig. 3] Because the real test micrographs are selected to have fractions matching the conditioning vectors, this is a selected qualitative illustration; please state the selection criterion or use randomly chosen test images.
  6. [Section 2.3] The sentence "equivalent to 5000 synthetic samples and 500 real samples (at 1:10 ratio)" is ambiguous about how many real samples are used at the other ratios and whether the same real subset is reused; please clarify the dataset construction.
  7. [Section 3.3, Fig. 12(b)] The text says the 4-D vectors are "reduced to 2-D using PCA and visualized using T-SNE"; T-SNE is not a dimensionality reduction in that phrasing, so clarify the two-step procedure.
  8. [Section 3.1, Table 2] The text says upscaling "raises mask ISimage," but the corresponding table column is ISmask; align the terminology across text and table.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the segmentation gain is measured on a held-out test set and the phase-fraction adherence check is an internal diagnostic, not a fitted prediction.

full rationale

The paper's central claim is an empirical augmentation result: PF-DiffSeg synthesizes image-mask pairs, and downstream segmentation models trained with those pairs are evaluated on 5 held-out MetalDAM micrographs. This is not circular because the reported MIoU is not defined by the generative model's inputs; it is measured on real test annotations. The phase-fraction adherence check (Fig. 4) compares target conditioning vectors c' with fractions computed from generated masks, but the training objective in Eq. (3) contains no term that directly minimizes a mask-fraction error, so the agreement is an empirical property of the learned denoiser, not an identity. The mixture-based sampling of c' from perturbed real fractions is a data-generation choice, and the resulting class-ratio changes in Fig. 12(a) are by design, but the downstream segmentation improvements are not derived from that design by equation. The cited baselines (SegGuidedDiff, DCGAN+Pix2PixHD, MetalDAM) are external works with no author overlap, and no load-bearing argument reduces to a self-citation. The passage 'only highest metrics were reported' indicates test-set hyperparameter selection, which is a statistical evaluation concern rather than a circular reduction: the reported MIoU is a maximum over ratios, but it is not equal to the ratio by construction.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities. It relies on hand-chosen generation hyperparameters (sigma, epsilon, mixture weights, lambda, DDIM steps) and on the domain assumption that global phase fractions are a sufficient conditioning signal for microstructure generation.

free parameters (5)
  • perturbation scale sigma = 0.02
    Hand-chosen in Eq. (7), Section 2.2.2, to jitter real phase fractions when sampling conditioning vectors.
  • perturbation floor epsilon = 0.01
    Hand-chosen small constant in Eq. (7) to keep fractions positive for rare classes.
  • oversampling mixture weights = 70% real, 20% MA-rich, 10% defect-rich
    Hand-chosen in Section 2.2.2 to rebalance the synthetic dataset toward rare phases.
  • loss weight lambda = 1
    Selected in Table A1 as the best of lambda = 0, 0.5, 1 on generative FID and IS metrics.
  • DDIM steps T = 50
    Chosen in Section 2.2.1 for faster sampling than full DDPM; no ablations shown for T.
assumptions (4)
  • standard math Standard DDPM and DDIM formulations apply unchanged when image and mask channels are concatenated.
    Section 2.2.1 relies on Eq. (1) and Eq. (2) from DDPM and DDIM without modification for the 4-channel input.
  • domain assumption A single global phase-fraction vector is sufficient conditioning to produce realistic, phase-controlled microstructures.
    The entire generation strategy in Section 2.2.1 assumes global composition controls local morphology; no spatial or shape priors are used.
  • domain assumption The 500 real training tiles are representative of the material's phase morphologies.
    Section 2.1 crops a small number of micrographs into tiles and assumes the diffusion model can learn the distribution from this small set.
  • domain assumption Synthetic pairs generated from the training distribution remain valid for training a segmentation model that generalizes to held-out test images.
    The augmentation pipeline in Section 2.3 assumes no harmful distribution shift between synthetic and real data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Phase-fraction guided denoising diffusion model for augmenting multiphase steel microstructure segmentation via micrograph image-mask pair synthesis." pith.science (2026). https://pith.science/paper/PSXOZZCW

@misc{pith2026250800896,
  author       = {Pith},
  title        = {Pith review of: Phase-fraction guided denoising diffusion model for augmenting multiphase steel microstructure segmentation via micrograph image-mask pair synthesis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PSXOZZCW}},
  note         = {Machine review of arXiv:2508.00896}
}
read the original abstract

The effectiveness of machine learning in metallographic microstructure segmentation is often constrained by the lack of human-annotated phase masks, particularly for rare or compositionally complex morphologies within the metal alloy. We introduce PF-DiffSeg, a phase-fraction controlled, one-stage denoising diffusion framework that jointly synthesizes microstructure images and their corresponding segmentation masks in a single generative trajectory to further improve segmentation accuracy. By conditioning on global phase-fraction vectors, augmented to represent real data distribution and emphasize minority classes, our model generates compositionally valid and structurally coherent microstructure image and mask samples that improve both data diversity and training efficiency. Evaluated on the MetalDAM benchmark for additively manufactured multiphase steel, our synthetic augmentation method yields notable improvements in segmentation accuracy compared to standard augmentation strategies especially in minority classes and further outperforms a two-stage mask-guided diffusion and generative adversarial network (GAN) baselines, while also reducing inference time compared to conventional approach. The method integrates generation and conditioning into a unified framework, offering a scalable solution for data augmentation in metallographic applications.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 25 canonical work pages

  1. [5]

    C. Shen, J. Zhao, M. Huang, C. Wang, Y . Zhang, W. Xu, S. Zheng, Generation of micrograph-annotation pairs for steel microstructure recognition using the hybrid deep generative model in the case of an extremely small and imbalanced dataset, Mater Charact 217 (2024) 114407. https://doi.org/10.1016/j.matchar.2024.114407

    work page arXiv 2024

  2. [1]

    Alrfou, T

    K. Alrfou, T. Zhao, A. Kordijazi, Deep Learning Methods for Microstructural Image Analysis: The State-of-the-Art and Future Perspectives, Integr Mater Manuf Innov 13 (2024) 703–731. https://doi.org/10.1007/s40192-024-00369-z

    work page doi:10.1007/s40192-024-00369-z 2024

  3. [2]

    DeCost, B

    B.L. DeCost, B. Lei, T. Francis, E.A. Holm, High Throughput Quantitative Metallography for Complex Microstructures Using Deep Learning: A Case Study in Ultrahigh Carbon Steel, Microscopy and Microanalysis 25 (2019) 21–29. https://doi.org/10.1017/S1431927618015635

    doi:10.1017/s1431927618015635 2019

  4. [3]

    Durmaz, M

    A.R. Durmaz, M. Müller, B. Lei, A. Thomas, D. Britz, E.A. Holm, C. Eberl, F. Mücklich, P. Gumbsch, A deep learning approach for complex microstructure inference, Nat Commun 12 (2021) 6272. https://doi.org/10.1038/s41467-021-26565-5

    work page doi:10.1038/s41467-021-26565-5 2021

  5. [4]

    Stuckner, B

    J. Stuckner, B. Harder, T.M. Smith, Microstructure segmentation with deep learning encoders pre-trained on a large microscopy dataset, NPJ Comput Mater 8 (2022) 200. https://doi.org/10.1038/s41524-022-00878-5

    doi:10.1038/s41524-022-00878-5 2022

  6. [6]

    Ragone, R

    M. Ragone, R. Shahabazian-Yassar, F. Mashayek, V . Yurkiv, Deep learning modeling in microscopy imaging: A review of materials science applications, Prog Mater Sci 138 (2023) 101165. https://doi.org/10.1016/j.pmatsci.2023.101165

    arXiv 2023

  7. [7]

    Luengo, R

    J. Luengo, R. Moreno, I. Sevillano, D. Charte, A. Peláez-Vegas, M. Fernández-Moreno, P. Mesejo, F. Herrera, A tutorial on the segmentation of metallographic images: Taxonomy, new MetalDAM dataset, deep learning-based ensemble model, experimental analysis and challenges, Information Fusion 78 (2022) 232–253. https://doi.org/10.1016/j.inffus.2021.09.018

    work page doi:10.1016/j.inffus.2021.09.018 2022

  8. [8]

    Biswas, R

    M. Biswas, R. Pramanik, S. Sen, A. Sinitca, D. Kaplun, R. Sarkar, Microstructural segmentation using a union of attention guided U-Net models with different color 29 transformed images, Sci Rep 13 (2023) 5737. https://doi.org/10.1038/s41598-023-32318- 9

    work page doi:10.1038/s41598-023-32318- 2023

Show all 35 references
  1. [9]

    X. Ma, Y . Yu, Training Tricks for Steel Microstructure Segmentation with Deep Learning, Processes (2023). https://doi.org/10.3390/pr11123298

    2023 doi

  2. [10]

    H. Kim, J. Inoue, T. Kasuya, Unsupervised microstructure segmentation by mimicking metallurgists’ approach to pattern recognition, Sci Rep 10 (2020) 17835. https://doi.org/10.1038/s41598-020-74935-8

    2020 doi

  3. [11]

    Na, S.-J

    J. Na, S.-J. Kim, H. Kim, S.-H. Kang, S. Lee, A unified microstructure segmentation approach via human-in-the-loop machine learning, Acta Mater 255 (2023) 119086. https://doi.org/10.1016/j.actamat.2023.119086

    2023

  4. [12]

    F. Chen, Y . Zhang, Y . Guo, Z. Liu, S. Du, Semi-Supervised Metallographic Image Segmentation via Consistency Regularization and Contrastive Learning, IEEE Access 11 (2023) 87398–87408. https://doi.org/10.1109/ACCESS.2023.3305269

    2023

  5. [13]

    Goodfellow, J

    I.J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y . Bengio, Generative Adversarial Networks, (2014)

    2014

  6. [14]

    Panda, R

    A. Panda, R. Naskar, S. Pal, Generative Adversarial Networks for Noise Removal in Plain Carbon Steel Microstructure Images, IEEE Sens Lett (2022). https://doi.org/10.1109/lsens.2022.3150776

    2022

  7. [15]

    Lee, H.J

    K.-H. Lee, H.J. Lim, G.J. Yun, A data-driven framework for designing microstructure of multifunctional composites with deep-learned diffusion-based generative models, Eng Appl Artif Intell (2024). https://doi.org/10.1016/j.engappai.2023.107590

    2024

  8. [16]

    Safiuddin, CH

    M. Safiuddin, CH. Likith Reddy, G. Vasantada, C.H.J.N.S. Harsha, S. Gangolu, Establishing Process-Structure Linkages Using Generative Adversarial Networks, in: 2025: pp. 497–509. https://doi.org/10.1007/978-981-97-6367-2_39

    2025 doi

  9. [17]

    Jangid, N.R

    D.K. Jangid, N.R. Brodnik, A. Khan, M.G. Goebel, M.P. Echlin, T.M. Pollock, S.H. Daly, B.S. Manjunath, 3D Grain Shape Generation in Polycrystals Using Generative Adversarial Networks, Integr Mater Manuf Innov 11 (2022) 71–84. https://doi.org/10.1007/s40192- 021-00244-1

    2022 doi

  10. [18]

    X. Li, S. Zhou, X. Liu, J. Zang, W. Fu, W. Lu, H. Zhang, Z. Yan, 3D microstructure reconstruction and characterization of porous materials using a cross-sectional SEM image and deep learning, Heliyon 10 (2024) e39185. https://doi.org/10.1016/j.heliyon.2024.e39185

    2024 doi

  11. [19]

    J. Ho, A. Jain, P. Abbeel, Denoising Diffusion Probabilistic Models, (2020)

    2020

  12. [20]

    Azqadan, H

    E. Azqadan, H. Jahed, A. Arami, Predictive microstructure image generation using denoising diffusion probabilistic models, Acta Mater 261 (2023) 119406. https://doi.org/10.1016/j.actamat.2023.119406

    2023

  13. [21]

    Lee, G.J

    K.-H. Lee, G.J. Yun, Microstructure reconstruction using diffusion-based generative models, (2022). 30

    2022

  14. [22]

    Fernandez-Zelaia, J

    P. Fernandez-Zelaia, J. Cheng, J. Mayeur, A.K. Ziabari, M.M. Kirka, Digital polycrystalline microstructure generation using diffusion probabilistic models, Materialia (Oxf) 33 (2024) 101976. https://doi.org/10.1016/j.mtla.2023.101976

    2024

  15. [23]

    Hoffman, C

    N. Hoffman, C. Diniz, D. Liu, T. Rodgers, A. Tran, M. Fuge, GrainPaint: A multi-scale diffusion-based generative model for microstructure reconstruction of large-scale objects, Acta Mater 288 (2025) 120784. https://doi.org/10.1016/j.actamat.2025.120784

    2025

  16. [24]

    N. Konz, Y . Chen, H. Dong, M.A. Mazurowski, Anatomically-Controllable Medical Image Generation with Segmentation-Guided Diffusion Models, (2024)

    2024

  17. [25]

    Sushko, D

    V . Sushko, D. Zhang, J. Gall, A. Khoreva, One-Shot Synthesis of Images and Segmentation Masks, (2022)

    2022

  18. [26]

    Toker, M

    A. Toker, M. Eisenberger, D. Cremers, L. Leal-Taixé, SatSynth: Augmenting Image-Mask Pairs through Diffusion Models for Aerial Semantic Segmentation, (2024)

    2024

  19. [27]

    Alimanov, M.B

    A. Alimanov, M.B. Islam, Denoising Diffusion Probabilistic Model for Retinal Image Generation and Segmentation, in: 2023 IEEE International Conference on Computational Photography (ICCP), IEEE, 2023: pp. 1–12. https://doi.org/10.1109/ICCP56744.2023.10233841

    2023

  20. [28]

    J. Song, C. Meng, S. Ermon, Denoising Diffusion Implicit Models, (2020)

    2020

  21. [29]

    Saharia, J

    C. Saharia, J. Ho, W. Chan, T. Salimans, D.J. Fleet, M. Norouzi, Image Super-Resolution via Iterative Refinement, (2021)

    2021

  22. [30]

    Ronneberger, P

    O. Ronneberger, P. Fischer, T. Brox, U-Net: Convolutional Networks for Biomedical Image Segmentation, (2015)

    2015

  23. [31]

    Zhou, M.M.R

    Z. Zhou, M.M.R. Siddiquee, N. Tajbakhsh, J. Liang, UNet++: A Nested U-Net Architecture for Medical Image Segmentation, (2018)

    2018

  24. [32]

    Chaurasia, E

    A. Chaurasia, E. Culurciello, LinkNet: Exploiting Encoder Representations for Efficient Semantic Segmentation, (2017). https://doi.org/10.1109/VCIP.2017.8305148

    2017

  25. [33]

    T. Fan, G. Wang, Y . Li, H. Wang, MA-Net: A Multi-Scale Attention Network for Liver and Tumor Segmentation, IEEE Access 8 (2020) 179656–179665. https://doi.org/10.1109/ACCESS.2020.3025372

    2020

  26. [34]

    K. He, X. Zhang, S. Ren, J. Sun, Deep Residual Learning for Image Recognition, (2015)

    2015

  27. [35]

    Rombach, A

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, B. Ommer, High-Resolution Image Synthesis with Latent Diffusion Models, (2022)

    2022

Pith tools

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