Pith. sign in

REVIEW 3 major objections 5 minor 28 references

ProBAG: Prototype-Guided Boundary-Aware Graph Diffusion for Weakly Supervised Histopathology Segmentation

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read ProBAG generates higher-quality tissue pseudo-masks from image-level labels alone, using hybrid text-visual prototypes and boundary-aware graph diffusion, and improves over prior weakly supervised histopathology segmentation methods by…

desk verdict The text-prototype result is solid, but the central claim about pseudo-mask quality is not actually tested. read the letter →

arxiv 2608.11765 v1 pith:JIBBDWKR submitted 2026-08-12 cs.CV

classification cs.CV
keywords weaklysupervisedsemanticsegmentationhistopathologytissueclassactivationmapstext-visualprototypesgraphdiffusionpathologyfoundationmodelspseudo-maskgenerationboundary-awarerefinement
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

ProBAG is a stage-1 pseudo-mask generator for weakly supervised histopathology segmentation: it produces dense tissue labels from image-level class labels alone, with no pixel-level supervision, no CRF, and no external segmentation model. The paper claims that matching dataset-specific visual prototypes and pathology-aligned CONCH text prototypes to multi-scale features from a frozen UNI encoder yields markedly better masks than recent weak supervision methods, and that two targeted corrections add further gains: an activation-balance operator that reshuffles class competition while preserving each pixel's total foreground mass, and a one-step graph diffusion whose feature affinities are penalized by an attention-context discrepancy from late transformer blocks. On BCSS-WSSS, ProBAG improves over the PBIP baseline by +4.13 mIoU, and on LUAD-HistoSeg it improves over the strongest retrained baseline by +2.15 mIoU. Ablations isolate the source of the gain: pathology-aligned text semantics contribute the largest share, while calibration and graph refinement contribute a smaller complementary correction. If the claim holds, it offers a cheaper path to training segmentation models for histopathology without expert pixel-level annotation.

What carries the argument

The carrying mechanism is the hybrid subclass prototype $p_k = \alpha \bar{t}_k + (1-\alpha)\bar{v}_k$, an $\ell^2$-normalized blend of a CONCH pathology text prototype $\bar{t}_k$ and a dataset-specific visual prototype $\bar{v}_k$, matched to multi-scale features from a frozen UNI encoder by scaled cosine similarity to produce class activation maps. Two corrections carry the argument. The activation-balance operator $\hat{S}_c(u) = m(u)\, (S_c(u)+\epsilon)^{\gamma_c} / \sum_j (S_j(u)+\epsilon)^{\gamma_j}$ keeps the per-pixel foreground mass $m(u)=\sum_j S_j(u)$ unchanged while reweighting which class claims it. Boundary-aware graph diffusion builds a graph on the deepest feature map with edge weights $W_{ij} \propto \exp(K_{ij}/\tau - \beta D_{ij})$, where $K_{ij}$ is feature affinity and $D_{ij}=\frac12\lVert a_i-a_j\rVert_1$ is the attention-context discrepancy between late-block self-attention profiles; subtracting $\beta D_{ij}$ lowers diffusion probability across contextually distinct tissue interfaces. The one-step residual update $z_c = (1-\lambda_c)s_c + \lambda_c W s_c$ preserves the original CAM evidence while applying a single constrained correction.

What would settle it

Run the released stage-1 pipeline on BCSS-WSSS with $\beta=0$ vs $\beta=2$ and evaluate with a boundary-sensitive metric such as boundary IoU or contour F1 on the four tissue classes; if the $\beta=2$ masks show no improvement over feature-only diffusion at interfaces despite $D_{ij}$ being large across known tissue boundaries, the attention-context discrepancy is not acting as a usable boundary proxy.

Watch

Extended reading notes

Core claim

The central discovery is that in weakly supervised histopathology segmentation, a prototype-guided boundary-aware graph diffusion pipeline can generate pseudo-masks whose quality approaches or exceeds the stage-2 segmentation of earlier methods, purely from image-level labels. The paper argues that the weakness of CAM-based localization in histology is two-fold: independently normalized class activations let broad or sharply activated tissues dominate pixel-wise competition, and feature-only affinity propagation leaks across tissue interfaces with similar local appearance. ProBAG addresses the first with class-wise power recalibration under an exact foreground-mass-preservation constraint, and the second with one-step graph diffusion regularized by the ℓ1 discrepancy between late-block self-attention profiles, used as a soft structural boundary cue rather than an explicit detector. The direct stage-1 masks reach 73.48% mIoU on BCSS-WSSS, above the 73.16% mIoU of the full two-stage system, and ablations show text-prototype semantics are the leading contributor while the boundary penalty yields a modest but consistent correction.

Load-bearing premise

The load-bearing premise is that the discrepancy between self-attention profiles at UNI block 23 is a trustworthy soft cue for where one tissue ends and another begins; the paper's own Table 4 shows region-based metrics move less than 0.2% mIoU when this penalty is varied from 0 to 3, so the boundary benefit is not yet statistically separated.

Editorial extensions

If this is right

  • Stage-1 pseudo-masks alone are competitive: on BCSS-WSSS they reach 73.48% mIoU without CRF or an external segmenter, suggesting the downstream stage-2 is not needed for the method's main benefit.
  • Pathology-aligned text prototypes are the dominant ingredient: adding CONCH text prototypes over a visual-only baseline raises mIoU by +4.18, so any practical WSSS pipeline for histopathology should incorporate domain-aligned text semantics.
  • The activation-balance operator decouples where foreground evidence lives from which class claims it, implying that class-imbalance problems in CAM fusion can be corrected without creating or suppressing evidence.
  • One-step graph diffusion with attention-context regularization yields a small but consistently positive correction (+0.43 mIoU combined with activation balance), indicating feature-only propagation can be improved without iterative smoothing that would erase small structures.
  • On a second dataset, LUAD-HistoSeg, ProBAG improves over the strongest retrained baseline by +2.15 mIoU, providing evidence the recipe transfers beyond BCSS.

Reading between the lines

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

  • Editorial inference: because Table 4 shows region metrics vary by less than 0.2% mIoU across $\beta\in[0,3]$, the attention-context penalty may be functioning mainly as a mild regularizer rather than a precise boundary detector; a boundary-sensitive metric such as contour F1 would likely reveal how much of the advertised boundary awareness is real.
  • Editorial inference: the fact that stage-1 masks (73.48 mIoU) end up slightly above the two-stage output (73.16 mIoU) suggests downstream training may be adding pseudo-label noise; a testable extension is to train several stage-2 decoders from the same masks and check whether any variant can preserve the stage-1 lead.
  • Editorial inference: the paper leaves cross-backbone comparisons open; a natural test is to swap the frozen UNI features for another pathology encoder while keeping CONCH text prototypes, predicting the text-prototype gain is encoder-independent if the cause is semantic alignment rather than feature quality.
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 / 5 minor

Summary. The paper proposes ProBAG, a stage-1 weakly supervised pseudo-mask generator for histopathology tissue segmentation. It combines dataset-specific visual prototypes with pathology-aligned CONCH text prototypes over multi-scale frozen UNI features, applies class-wise power recalibration that preserves each pixel's total foreground activation mass, and refines the masks via one-step graph diffusion whose affinities are penalized by a late-transformer attention-context discrepancy used as a soft boundary cue. Stage-1 masks are evaluated directly in internal ablations on BCSS-WSSS; for comparison with two-stage WSSS systems, the masks supervise a Phikon-FPN with LoRA on BCSS-WSSS and LUAD-HistoSeg. The main reported gains come from text prototypes (+4.18 mIoU), with activation balance and graph diffusion adding a further +0.43 mIoU.

Significance. If the central claim is established, ProBAG would provide a useful, CRF-free stage-1 pseudo-mask generator for histopathology WSSS, with a clean separation between stage-1 mask quality and stage-2 evaluation. The paper's strengths include a released codebase, unusually candid and explicit limitation statements, a controlled text-encoder comparison, and a clear formulation of the mass-preserving recalibration operator. However, the significance as framed—that ProBAG's pseudo-masks are higher quality than recent WSSS methods—is not yet directly evidenced, because the headline comparison conflates pseudo-mask quality with downstream architecture choices, and the boundary-aware component's effect is not statistically separable on region metrics. The contribution is potentially valuable, but the evaluation needs strengthening before the claimed significance is supported.

major comments (3)
  1. [Sec. 3.3, Table 1; Sec. 2 (stage-1 vs stage-2)] The load-bearing claim that ProBAG is superior as a stage-1 pseudo-mask generator is not directly tested. Table 1 compares complete two-stage systems: ProBAG's masks supervise a Phikon-FPN with LoRA, while baselines use their official downstream implementations with different foundation/backbone settings. The paper itself acknowledges this in the Table 1 note and Sec. 3.5. The stage-1 ablations in Tables 2–4 compare only internal ProBAG variants; they do not evaluate PBIP, TPRO, Proto2Seg, or CAM under the same stage-1 protocol. I request a stage-1-only comparison in which all methods generate pseudo-masks from the same frozen features and the same evaluation pipeline, with at least a common backbone, and report pseudo-mask mIoU. Without this, the headline +4.13 and +2.15 mIoU margins could be attributed to the Phikon+LoRA stage-2 segmenter or the UNI/CONCH foundation stack rather than to the proposed stage-1 design.
  2. [Sec. 3.4, Table 4] The attention-context boundary penalty β, which is one of the two named complementary mechanisms, has no statistically separable effect on the reported region metrics: increasing β from 0 to 2 changes mIoU by 0.09, which is below the reported standard deviation of roughly 0.2. The paper states this limitation, but it means the boundary-aware claim is currently supported only by qualitative Figure 2. Please add a boundary-sensitive metric (e.g., boundary IoU or contour-based F-measure) and report β=0 versus β>0 on that metric, or explicitly reposition BAGD as a regularization whose benefit is not demonstrated in the present experiments.
  3. [Sec. 3.3, Table 1] Table 1 reports single-run results for the final systems. Given that the stage-1 ablations report standard deviations around 0.2 mIoU and the BAGD gain is below that level, single-run margins on the headline comparison are not sufficient to establish 'consistent gains.' Please report mean±std over at least three seeds for ProBAG and for the strongest baseline (PBIP on BCSS-WSSS, CAM on LUAD-HistoSeg) on both datasets, so that the main comparison has error bars.
minor comments (5)
  1. [Author affiliations] The author list uses superscripts 6 and 7 for Tam Tran and Zhi Huang, but the affiliation list only enumerates affiliations 1–4; please fix the numbering or the affiliation markers.
  2. [Sec. 2.1, Eq. (4)] The parenthetical text 'the implementation option named "max"' is confusing because the described operation is a softmax-weighted sum, not a maximum; consider renaming this option or removing the parenthetical.
  3. [Fig. 2 caption] The sentence 'Figure 2 Figure 2 visually suggests' duplicates 'Figure 2'; please correct the duplication.
  4. [Sec. 2.2, Eq. (8)] Since the exact preservation of the foreground mass is a stated contribution, please add one sentence or a short derivation showing that summing bS_c over c recovers m(u), to make the property visually evident from the equation.
  5. [Sec. 3.3, Table 1 footnote] PBIP's LUAD results are marked with † and reflect stage-1 pseudo-mask evaluation, while ProBAG's LUAD row is a two-stage result; this makes the LUAD comparison asymmetric, so please clarify in the caption or text how the '+2.15 mIoU over CAM' margin should be interpreted.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation is self-contained and the evaluation rests on external benchmarks.

full rationale

The paper's derivation chain is not circular. Prototype CAMs (Eqs. 2-4) are trained from image-level labels; activation balance (Eq. 8) is a fixed mass-preserving operator; graph diffusion (Eqs. 9-12) uses frozen UNI affinities and attention profiles with fixed hyperparameters and no pixel supervision. Pseudo-masks from Eq. 13 are evaluated directly in Tables 2-4 and via a separate Phikon-FPN in Table 1 against external benchmarks. No equation reduces to a fitted target or to the paper's own prior output. The only self-citation (ref. [22], UniSemAlign) appears in a motivation sentence ('Recent studies have also demonstrated the utility of pathology foundation models for histopathology segmentation [9,22]') and is not used to justify any equation, parameter choice, or uniqueness claim. The paper explicitly acknowledges that Table 1 compares complete systems with different backbones and single runs; that is a limitation on the strength of the empirical claim, not a circular derivation. Therefore no circularity step is present.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard WSSS assumptions and domain assumptions about pathology foundation models. No novel physical or conceptual entities are introduced. The main unaccounted tuning choices are the unreported gamma_c exponents, the per-class lambda_c set, and the attention block selection, all of which affect the final pseudo-masks.

free parameters (6)
  • alpha (hybrid text-visual weight) = 0.6
    Fixed in Section 2.1; controls the blend between CONCH text prototypes and visual prototypes, directly shaping the CAM logits in Eqs. 2-3.
  • per-class diffusion weight lambda_c = (0.35, 0.35, 0.20, 0.45) for (TUM, STR, LYM, NEC)
    Chosen in the released BCSS configuration (Section 2.3) and used in Eq. 12; interpolates between the original CAM and the diffused map.
  • graph resolution = 7x7
    Graph built at 7x7 resolution from F4 (Section 2.3); downsampling and upsampling of CAMs depends on this choice.
  • delta, tau, beta = 0.55, 0.10, 2
    Kernel threshold, temperature, and boundary penalty in Eqs. 9-11; selected on validation, with beta shown to be insensitive in Table 4.
  • gamma_c (power recalibration exponents) = not reported
    Appear in Eq. 8 as class-wise exponents; no values are given in the paper, so the recalibration is not fully reproducible.
  • attention block for context profile = 23
    The attention-context discrepancy is extracted from late UNI block 23 (Section 2.3); a different block changes Dij and thus the diffusion weights.
assumptions (4)
  • domain assumption Frozen UNI and CONCH are pathology-aligned and provide useful dense features and text semantics for tissue segmentation.
    Invoked in Sections 1 and 2.1; the paper explicitly notes it does not provide a controlled comparison against alternative pathology foundation models (Section 3.5).
  • domain assumption Image-level multi-label annotations suffice to learn pseudo-masks.
    Standard WSSS premise; the entire method relies on BCE classification from image-level labels and does not use pixel labels in training.
  • ad hoc to paper The attention-context discrepancy Dij is a valid soft proxy for tissue boundaries.
    Introduced in Section 2.3, Eq. 10; Table 4 shows region metrics vary by less than 0.2% mIoU across beta, so this assumption is only weakly supported.
  • ad hoc to paper Softmax-weighted subclass aggregation (the implementation option named 'max') and parent-level merging are valid design choices.
    Section 2.1 describes this implementation choice for pseudo-mask generation without further justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ProBAG: Prototype-Guided Boundary-Aware Graph Diffusion for Weakly Supervised Histopathology Segmentation." pith.science (2026). https://pith.science/paper/JIBBDWKR

@misc{pith2026260811765,
  author       = {Pith},
  title        = {Pith review of: ProBAG: Prototype-Guided Boundary-Aware Graph Diffusion for Weakly Supervised Histopathology Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JIBBDWKR}},
  note         = {Machine review of arXiv:2608.11765}
}
read the original abstract

Weakly supervised semantic segmentation enables histopathology tissue segmentation from image-level annotations, avoiding costly pixel-level labeling by expert pathologists. However, CAM-based methods often localize only highly discriminative regions and remain unreliable near tissue interfaces. We propose ProBAG, a stage-1 pseudo-mask generator that combines dataset-specific visual prototypes with pathology-aligned CONCH text prototypes over multi-scale frozen UNI features. ProBAG introduces two complementary mechanisms: class-wise power recalibration that reshapes inter-class competition while preserving the total foreground activation mass at each pixel, and one-step graph diffusion in which feature affinities are penalized by a late-transformer attention-context discrepancy used as a soft structural boundary cue. The resulting stage-1 pseudo-masks require neither CRF nor an external segmentation model; for complete two-stage comparison, they additionally supervise a downstream Phikon-FPN segmenter. Experiments on BCSS-WSSS and LUAD-HistoSeg show consistent gains over recent WSSS approaches, while ablations indicate that pathology-aligned text semantics provide the largest improvement and graph refinement provides a smaller complementary gain. The code is available at: https://github.com/wterrr/WSSS

Figures

Figures reproduced from arXiv: 2608.11765 by the authors.

Figure 1
Figure 1. Overview of the ProBAG framework. Hybrid text–visual prototypes are matched with multi-scale features from a frozen pathology encoder to generate tis￾sue CAMs, which are then fused, activation-balanced, and refined by boundary-aware graph diffusion to produce pseudo masks [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Qualitative comparison of final segmentation results on BCSS-WSSS and LUAD-HistoSeg test patches. GT denotes the ground truth segmentation. by +0.99%. This controlled comparison supports the use of pathology-aligned text semantics, while not constituting a broader comparison of visual foundation backbones [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 11 canonical work pages

  1. [1]

    Ahn, J., Cho, S., Kwak, S.: Weakly supervised learning of instance segmentation with inter-pixel relations (2019),https://arxiv.org/abs/1904.05044

  2. [2]

    Ahn, J., Kwak, S.: Learning pixel-level semantic affinity with image-level super- vision for weakly supervised semantic segmentation (2018),https://arxiv.org/ abs/1803.10464

  3. [3]

    Amgad, M., Elfandy, H., Hussein, H., Atteya, L.A., Elsebaie, M.A.T., Abo Elnasr, L.S., Sakr, R.A., Salem, H.S.E., Ismail, A.F., Saad, A.M., Ahmed, J., Elsebaie, M.A.T., Rahman, M., Ruhban, I.A., Elgazar, N.M., Alagha, Y., Osman, M.H., Alhusseiny, A.M., Khalaf, M.M., Younes, A.A.F., Abdulkarim, A., Younes, D.M., Gadallah, A.M., Elkashash, A.M., Fala, S.Y.,...

  4. [4]

    Nature medicine25, 1301 – 1309 (2019),https://api.semanticscholar.org/ CorpusID:196814162

    Campanella, G., Hanna, M.G., Geneslaw, L., Miraflor, A.P., Silva, V.W.K., Busam, K.J., Brogi, E., Reuter, V.E., Klimstra, D.S., Fuchs, T.J.: Clinical-grade computa- tional pathology using weakly supervised deep learning on whole slide images. Nature medicine25, 1301 – 1309 (2019),https://api.semanticscholar.org/ CorpusID:196814162

  5. [5]

    Nature Medicine (2024)

    Chen, R.J., Ding, T., Lu, M.Y., Williamson, D.F., Jaume, G., Chen, B., Zhang, A., Shao, D., Song, A.H., Shaban, M., et al.: Towards a general-purpose foundation model for computational pathology. Nature Medicine (2024)

  6. [6]

    Chen, Z., Wang, T., Wu, X., Hua, X.S., Zhang, H., Sun, Q.: Class re-activation maps for weakly-supervised semantic segmentation (2022),https://arxiv.org/ abs/2203.00962

  7. [7]

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., Houlsby, N.: An image is worth 16x16 words: Transformers for image recognition at scale (2021), https://arxiv.org/abs/2010.11929

  8. [8]

    In: AAAI Conference on Artificial Intelligence (2023), https://api.semanticscholar.org/CorpusID:259735474

    Fang, Z., Chen, Y., Wang, Y., Wang, Z., Ji, X., Zhang, Y.: Weakly-supervised se- mantic segmentation for histopathology images based on dataset synthesis and fea- ture consistency constraint. In: AAAI Conference on Artificial Intelligence (2023), https://api.semanticscholar.org/CorpusID:259735474

Show all 28 references
  1. [9]

    Fu, M., Fu, F., Ling, X., Yuan, H., Guan, T., He, Y., Zhu, L.: Multimodal prototype alignment for semi-supervised pathology image segmentation (2025), https://arxiv.org/abs/2508.19574

  2. [10]

    Han, C., Pan, X., Yan, L., Lin, H., Li, B., Yao, S., Lv, S., Shi, Z., Mai, J., Lin, J., Zhao, B., Xu, Z., Wang, Z., Wang, Y., Zhang, Y., Wang, H., Zhu, C., Lin, C., Mao, L., Wu, M., Duan, L., Zhu, J., Hu, D., Fang, Z., Chen, Y., Zhang, Y., Li, Y., Zou, Y., Yu, Y., Li, X., Li, ...

  3. [11]

    Kipf, T.N., Welling, M.: Semi-supervised classification with graph convolutional networks (2017),https://arxiv.org/abs/1609.02907

  4. [12]

    Kolesnikov, A., Lampert, C.H.: Seed, expand and constrain: Three principles for weakly-supervised image segmentation (2016),https://arxiv.org/abs/1603. 06098

  5. [13]

    Lee, J., Oh, S.J., Yun, S., Choe, J., Kim, E., Yoon, S.: Weakly supervised seman- tic segmentation using out-of-distribution data (2022),https://arxiv.org/abs/ 2203.03860

  6. [14]

    Lin,T.Y.,Dollár,P.,Girshick,R.,He,K.,Hariharan,B.,Belongie,S.:Featurepyra- mid networks for object detection (2017),https://arxiv.org/abs/1612.03144

  7. [15]

    Nature Medicine30, 863–874 (2024)

    Lu, M.Y., Chen, B., Williamson, D.F., Chen, R.J., Liang, I., Ding, T., Jaume, G., Odintsov, I., Le, L.P., Gerber, G., et al.: A visual-language foundation model for computational pathology. Nature Medicine30, 863–874 (2024)

  8. [16]

    Minaee, S., Boykov, Y., Porikli, F., Plaza, A., Kehtarnavaz, N., Terzopoulos, D.: Image segmentation using deep learning: A survey (2020),https://arxiv.org/ abs/2001.05566

  9. [17]

    van den Oord, A., Li, Y., Vinyals, O.: Representation learning with contrastive predictive coding (2019),https://arxiv.org/abs/1807.03748

  10. [18]

    Pan, W., Yan, J., Chen, H., Yang, J., Xu, Z., Li, X., Yao, J.: Human-machine interactive tissue prototype learning for label-efficient histopathology image seg- mentation (2023),https://arxiv.org/abs/2211.14491 12 Duy-Dong Nguyen et al

  11. [19]

    Ru, L., Zhan, Y., Yu, B., Du, B.: Learning affinity from attention: End-to- end weakly-supervised semantic segmentation with transformers (2022),https: //arxiv.org/abs/2203.02664

  12. [20]

    International Journal of Computer Vision128(2), 336–359 (Oct 2019)

    Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., Batra, D.: Grad-cam: Visual explanations from deep networks via gradient-based localiza- tion. International Journal of Computer Vision128(2), 336–359 (Oct 2019). https://doi.org/10.1007/s11263-019-01228-7,http:...

  13. [21]

    org/abs/2503.12068

    Tang, Q., Fan, L., Pagnucco, M., Song, Y.: Prototype-based image prompting for weakly supervised histopathological image segmentation (2025),https://arxiv. org/abs/2503.12068

  14. [22]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops

    Van Thai, L., Nguyen, T.D., Pham, H.N., Thi, L.A.D., Nguyen, D.D., Bui, N.L.Q.: Unisemalign: Text-prototype alignment with a foundation encoder for semi-supervised histopathology segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognitio...

  15. [23]

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, L., Polosukhin, I.: Attention is all you need (2023),https://arxiv.org/abs/1706. 03762

  16. [24]

    Wang, Y., Zhang, J., Kan, M., Shan, S., Chen, X.: Self-supervised equivariant attention mechanism for weakly supervised semantic segmentation (2020),https: //arxiv.org/abs/2004.04581

  17. [25]

    Wang, Z., Wu, Z., Agarwal, D., Sun, J.: Medclip: Contrastive learning from un- paired medical images and text (2022),https://arxiv.org/abs/2210.10163

  18. [26]

    Xie, J., Hou, X., Ye, K., Shen, L.: Cross language image matching for weakly supervised semantic segmentation (2022),https://arxiv.org/abs/2203.02668

  19. [27]

    In: International Conference on Medi- cal Image Computing and Computer-Assisted Intervention (2023),https://api

    Zhang, S., Zhang, J., Xie, Y., Xia, Y.: Tpro: Text-prompting-based weakly super- vised histopathology tissue segmentation. In: International Conference on Medi- cal Image Computing and Computer-Assisted Intervention (2023),https://api. semanticscholar.org/CorpusID:263673303

  20. [28]

    Zhou, B., Khosla, A., Lapedriza, A., Oliva, A., Torralba, A.: Learning deep features for discriminative localization (2015),https://arxiv.org/abs/1512.04150

Pith tools

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