Pith. sign in

REVIEW 3 major objections 5 minor 57 references

Slender Object Scene Segmentation in Remote Sensing Image Based on Learnable Morphological Skeleton with Segment Anything Model

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A learnable morphological skeleton prior embedded in SAM's decoder lifts recall by about 2 percent on slender remote-sensing objects.

desk verdict A clever variational unrolling of a smooth morphological skeleton into SAM, with consistent but modest gains whose origin in the skeleton prior is not yet convincingly established. read the letter →

arxiv 2411.08592 v2 pith:TGK5Q6AO submitted 2024-11-13 cs.CV

Jun Xie , Wenxiao Li , Faqiang Wang , Liqiang Zhang , Zhengyang Hou , Jun Liu This is my paper · ORCID
classification cs.CV MSC 68U1068T07
keywords semanticsegmentationremotesensingmorphologicalskeletonSegmentAnythingModelvariationalmethodsoftthresholddynamicsalgorithmunrollingslenderobject
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that the fine-detail loss seen in encoder-decoder segmentation models can be repaired by injecting a morphological skeleton prior at the decoding stage, rather than only into the loss function. It constructs a differentiable approximation of the morphological skeleton and a matching variational segmentation model, then unrolls the solver into a module called MorSP that plugs into SAM. The reported result is consistent gains on building, road, and water datasets, roughly 2 percent higher recall than SAM fine-tuned normally, plus better robustness to Gaussian noise. If true, this gives a mathematically interpretable way to keep structural priors active during prediction, not just during training.

What carries the argument

The load-bearing device is the smooth morphological skeleton $S^\alpha(u)$, obtained by replacing the non-differentiable max and min of dilation and erosion with log-sum-exp soft operators, together with a variational segmentation energy that appends a skeleton-matching cost $C(u)=\frac{1}{2}\int_\Omega |S^\alpha(u)(x)-S^\alpha(g)(x)|^2\,dx$ to the soft threshold dynamics model. Operator splitting and $L^1$ duality turn the minimization into three subproblems, the $u$-update of which is a sigmoid; unrolling roughly twenty iterations of that update, with a learnable skeleton token standing in for the unknown ground-truth skeleton, gives the MorSP module.

What would settle it

A reader could settle the mechanism by freezing all SAM-MorSP weights and replacing the learned skeleton token with a constant or with the skeleton of a different object class on the Massachusetts road test set: if road recall stays near the reported 74.33 while the skeleton input carries no road structure, the claimed skeleton-prior effect is not what produces the gain.

Watch

Extended reading notes

Core claim

The paper's central claim is that a smooth, learnable version of the morphological skeleton can be written as a variational regularizer and unrolled into a neural module, and that inserting this module into SAM's mask decoder makes the network preserve slender and topologically complex structures in remote sensing images. The integration works by adding a skeleton token to the mask decoder and replacing SAM's final thresholding with a learnable sigmoid solve of a variational model. On the reported building, road, and water benchmarks, SAM-MorSP achieves the best F1, IoU, precision, and recall among the compared fine-tuning strategies, with about a 2 percent recall gain over SAM fine-tuned without the module, and it degrades less under added Gaussian noise.

Load-bearing premise

The gain is credited to the skeleton regularizer, but the 20-step approximate solver is never shown to converge to the variational model's minimizer, so the improvement could in principle come from the extra learned parameters rather than from the skeleton prior.

Editorial extensions

If this is right

  • SAM-MorSP should segment narrow and elongated objects such as road intersections, roundabouts, and slender rivers more completely than SAM fine-tuned with plain cross-entropy or skeleton loss.
  • Because MorSP operates at the final decoding layer, it can be attached to other segmentation backbones; the paper demonstrates this with BEDSN and reports improved F1 and IoU on building and road datasets.
  • The model should degrade more gracefully under Gaussian noise than loss-only skeleton supervision, since the soft threshold dynamics regularization and smooth morphological operators act during inference as well as training.
  • Fine-tuning only about 4 percent of SAM ViT-B's parameters, through LoRA and MorSP, is enough to exceed full-decoder fine-tuning on the tested remote sensing benchmarks.

Reading between the lines

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

  • A direct test the paper does not report is ablating the skeleton token: if feeding a constant or random token preserves the reported recall gains, then the skeleton prior itself is not carrying the improvement.
  • The same module could transfer to non-remote-sensing line-structure tasks such as retinal vessel or power-line segmentation, where connectivity matters more than boundary accuracy; the paper only evaluates buildings, roads, and water.
  • Because the skeleton is learned from a token inside the decoder, it could in principle be conditioned on the prompt (point or box) to emphasize different topological structures, an extension the paper does not explore.
  • The paper offers no convergence analysis for its one-step approximate subproblem updates, so a controlled experiment separating the skeleton branch from the extra learned parameters would clarify whether the gains come from the variational prior or simply from more capacity.
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 SAM-MorSP, an extension of the Segment Anything Model that integrates a learnable morphological skeleton prior into the segmentation decoder. The authors derive a smooth approximation of morphological dilation/erosion and of the morphological skeleton, use it to build the variational segmentation model in Eq. (10) with an entropy term, a threshold-dynamics regularization, and a skeleton cost, and then unroll an alternating optimization scheme into a trainable MorSP module with a soft skeleton token. The module replaces SAM's final Heaviside decoding step. Experiments on building, road, and water datasets compare SAM-MorSP with SAM-fine and SAM-cl, reporting improved F1 and recall in most settings, including under Gaussian noise, and an additional experiment integrates MorSP into the BEDSN backbone.

Significance. If the claims hold, the paper offers a principled way to inject a topological/skeleton prior into a foundation model through a variational energy rather than only through a loss function. The derivations in Propositions 1-2 provide a clean log-sum-exp representation of morphological operators, and the proposed module is architecture-agnostic, as demonstrated by the BEDSN experiment. The noise-robustness study is a useful addition, and the comparison with a cl-dice-loss baseline helps separate the effect of the variational module from the skeleton loss. However, the interpretability claim depends on two points that are not established: the smoothness of the derived skeleton variation, and the claim that the truncated, approximate unrolled iterations actually solve the variational model. The empirical evidence also lacks error bars and has precision drops on several datasets, so the significance is conditional on these points being resolved.

major comments (3)
  1. [Section III-A, Proposition 3] Proposition 3 expresses the variation of S^α(u) using the Heaviside step function H, and S^α itself is defined through ReLU projections. Consequently, S^α is not smooth in u, and the displayed formula is not the derivative of a smooth function; this contradicts the paper's assertion that the proposed skeletal representation is smooth and therefore suitable for gradient backpropagation. Since δC/δu in Eq. (9) and the w-update in Algorithm 1 depend on this formula, the authors should either reinterpret the expression as a subgradient/formal derivative, or replace the ReLU and Heaviside functions with genuinely smooth approximations and re-derive the variation.
  2. [Section III-B and Algorithm 1] The paper states that the q and w subproblems are solved 'approximately' with one iteration each and that T = 20 layers are taken 'to save computational sources,' but it provides no convergence, monotonicity, or fixed-point analysis for the alternating scheme, and no numerical comparison with a converged solution of Eq. (10). As written, the MorSP forward pass may be far from the minimizer of the variational model, so the reported gains cannot be confidently attributed to the skeleton prior rather than to added trainable capacity in the soft skeleton token, the learnable hyperparameters, or the cl-dice loss. I ask for either a convergence/consistency analysis of Algorithm 1 or a numerical study reporting objective values and segmentation outputs for larger T and for a converged solution.
  3. [Section V-C and Tables III-VII] The headline claim of 'superior performance across all datasets, with an approximate 2% improvement in recall compared to SAM-fine' is not uniformly supported by the tables: on clean test sets, F1 gains over SAM-fine range from +0.79 to +1.45, recall changes range from +0.17 to +5.75, and precision drops on LAIS building (83.39 to 80.10), Massachusetts road (79.27 to 77.60), and WHU-OPT-SAR water (79.22 to 72.03). No error bars or statistical significance tests are reported, and Table V does not state whether the λl selection is made on the validation or test split; if the test set is used for hyperparameter selection, the reported numbers are optimistic. Please provide multi-run statistics and clarify or avoid test-set selection.
minor comments (5)
  1. [Table III footnote] The footnote says the three models have the same number of learnable parameters, but SAM-MorSP adds a soft skeleton token and learnable hyperparameters (γ, λ, α, η, ι) in addition to the modified decoder; please report the exact learnable-parameter count for each baseline.
  2. [Section II-A] There are several typos and grammatical errors, e.g., 'divid' should be 'divide', 'searve' should be 'serve', 'konwledge' should be 'knowledge', 'segementing' should be 'segmenting', and 'an 32GB' should be 'a 32GB'; a careful proofreading pass is needed.
  3. [Section V-D] The loss-weight experiment in Table V is reported on the Massachusetts road dataset, but the text does not state whether these numbers are on the validation set or the test set; this should be clarified to assess whether the hyperparameter was selected legitimately.
  4. [Section V-E] The computational-cost table reports 'Gflops' and inference time, but the hardware and measurement details are minimal; please state the GPU model, batch size, and whether inference time is measured on a single 1024×1024 image.
  5. [Section VI] The BEDSN experiment is described briefly without training details or error bars; adding the same experimental protocol as the SAM experiments would make the generalization claim more convincing.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the morphological skeleton is an imposed regularizer, the unrolled module is trained and tested against external baselines, and the cited STD derivation is independent support.

full rationale

The paper's derivation chain is self-contained rather than circular. The smooth dilation/erosion operators and the skeleton fidelity term C(u) in Eq. (9) are constructed from explicit mathematical propositions (Propositions 1-3) before being inserted as a regularizer into the variational model (10); the skeleton prior is an input constraint, not a quantity derived from the network output. The unrolled MorSP forward pass solves an explicit min-max problem (Algorithm 1 and Eq. (11)), and the soft skeleton token v is learned from training data under cl-dice supervision and then used at inference; no reported quantity reduces to a fitted constant or to the ground-truth skeleton by construction. The empirical comparison is against SAM-fine and SAM-cl on external benchmarks, with the cl-dice loss baseline controlling for the loss component. The cited STD model [27] is a published, parameter-free variational construction, and the paper reproduces the sigmoid u-update rather than assuming the paper's own target result; the overlap of author Jun Liu with [27] is therefore not load-bearing circularity. Section IV's oracle experiment using ground-truth skeleton Sα(g) is explicitly labeled as such and is not presented as the final prediction result. The T=20 truncation and approximate one-step q/w updates are a convergence and computational-accuracy concern, not a circularity, and the paper openly labels them as approximations. No uniqueness theorem or ansatz is smuggled via self-citation, and no known result is merely renamed. Under the quoted-evidence standard, no circular step is established.

Assumptions & free parameters 8 free parameters · 6 assumptions · 0 invented entities

The central claim rests on several learnable hyperparameters (gamma, lambda, alpha, eta, iota), a hand-chosen kernel size k, a loss weight selected on the test set, and a learned soft skeleton token. The axioms include standard smooth-approximation assumptions, the prior STD model, and a non-trivial assumption that the unrolled finite-horizon optimization approximates the stated variational model. No new physical entities are introduced.

free parameters (8)
  • gamma (entropy weight) = learnable, init 1
    Weight of the entropy regularization in the variational model (10); learnable during fine-tuning, initial value in Table I.
  • lambda (TD regularization weight) = learnable, init 1
    Weight of the threshold dynamics boundary regularization R(u) in (10); learnable, initial value in Table I.
  • alpha (smooth morphological temperature) = learnable, init 0.05
    Controls the log-sum-exp smoothness of dilation/erosion; small alpha approximates classical morphology. Learnable, initial value in Table I. If alpha grows, the smooth skeleton deviates from the true morphological skeleton.
  • eta (L1 penalty weight) = learnable, init 1
    Penalty weight for the consensus constraint w=u in the operator splitting; learnable, initial value in Table I.
  • iota (w-subproblem step size) = learnable, init 0.01
    Step size for the projected gradient update of w; learnable, initial value in Table I.
  • lambda_l (loss weight for cl-dice) = 0.1
    Weight balancing CE and skeleton loss; selected by tuning on the Massachusetts road test set (Section V-D, Table V), a test-set selection.
  • soft skeleton token v = learned vector
    The decoder's extra output token serves as the target skeleton in the MorSP module during inference; trained with cl-dice loss against ground-truth skeletons.
  • structuring element size k = 5
    Kernel size of the structuring element B in the smooth morphological operators; chosen by hand (Table I).
assumptions (6)
  • domain assumption The smooth dilation D_alpha via log-sum-exp converges to the morphological max operator as alpha approaches 0 for continuous images (Prop. 1).
    Requires u continuous in the structuring element neighborhood; for discrete images, convergence is approximate and depends on alpha.
  • domain assumption The threshold dynamics energy R(u)=lambda⟨u, f*(1-u)⟩ approximates boundary length (Eq. 2), taken from Liu et al. [27].
    The TD approximation is a known model from prior work by the same group; the paper relies on its validity.
  • ad hoc to paper The variational model (10) with entropy, TD, and skeleton prior is a suitable segmentation objective.
    The model is constructed in this paper; its well-posedness and minimizer properties are not analyzed.
  • ad hoc to paper The unrolled MorSP module with T=20 iterations and one approximate gradient step per subproblem approximately solves the variational model (10).
    No convergence analysis is provided; the paper states the approximate updates are solved with 'one iteration' (Section III-B) and T=20 is chosen for computational reasons (Section III-C). This is the weakest link.
  • domain assumption The learned soft skeleton token v at test time approximates the ground-truth skeleton well enough to serve as a useful prior.
    v is trained with cl-dice loss; at inference there is no guarantee it resembles the true skeleton, and the paper does not analyze failure cases.
  • domain assumption Labels in LAIS and Massachusetts datasets, despite known offsets and errors, are reliable enough to evaluate the claimed improvements.
    The paper notes label noise in both datasets but uses them as ground truth; noise could affect the relative ranking of methods.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Slender Object Scene Segmentation in Remote Sensing Image Based on Learnable Morphological Skeleton with Segment Anything Model." pith.science (2026). https://pith.science/paper/TGK5Q6AO

@misc{pith2026241108592,
  author       = {Pith},
  title        = {Pith review of: Slender Object Scene Segmentation in Remote Sensing Image Based on Learnable Morphological Skeleton with Segment Anything Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TGK5Q6AO}},
  note         = {Machine review of arXiv:2411.08592}
}
read the original abstract

Morphological methods play a crucial role in remote sensing image processing, due to their ability to capture and preserve small structural details. However, most of the existing deep learning models for semantic segmentation are based on the encoder-decoder architecture including U-net and Segment Anything Model (SAM), where the downsampling process tends to discard fine details. In this paper, we propose a new approach that integrates learnable morphological skeleton prior into deep neural networks using the variational method. To address the difficulty in backpropagation in neural networks caused by the non-differentiability presented in classical morphological operations, we provide a smooth representation of the morphological skeleton and design a variational segmentation model integrating morphological skeleton prior by employing operator splitting and dual methods. Then, we integrate this model into the network architecture of SAM, which is achieved by adding a token to mask decoder and modifying the final sigmoid layer, ensuring the final segmentation results preserve the skeleton structure as much as possible. Experimental results on remote sensing datasets, including buildings, roads and water, demonstrate that our method outperforms the original SAM on slender object segmentation and exhibits better generalization capability.

Figures

Figures reproduced from arXiv: 2411.08592 by the authors.

Figure 1
Figure 1. Illustration of skeleton extraction: from left to right are original image [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The network architecture of the proposed Morphological Skeleton [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. The performance of proposed MorSP module: the first column is the [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The network architecture of SAM-MorSP. The results are shown in [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Visual result of SAM-fine, SAM-cl, SAM-MorSP on WHU and LAIS building test sets. [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Visual result of SAM-fine, SAM-cl, SAM-MorSP on LAIS and Massachusetts roads test sets. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Visual result of SAM-fine, SAM-cl, SAM-MorSP on LAIS and WHU-OPT-SAR test set. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: Visual result of noise test. The first row (from left to right): noisy [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Visual result of BEDSN on Massachusetts road dataset. [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 51 canonical work pages

  1. [1]

    Some generalized order-disorder transformations,

    R. B. Potts, “Some generalized order-disorder transformations,” Math. Proc. Cambridge Philos. Soc. , vol. 48, no. 1, pp. 106–109, Jan 1952

    work page 1952

  2. [2]

    Snakes: Active contour models,

    M. Kass, A. Witkin, and D. Terzopoulos, “Snakes: Active contour models,” Int. J. Comput. Vis. , vol. 1, no. 4, pp. 321–331, Jan. 1988

    work page 1988

  3. [3]

    Fully convolutional networks for semantic segmentation,

    J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR) , Jun. 2015, pp. 3431–3440

    work page 2015

  4. [4]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Proc. 18th Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. (MICCAI). , May. 2015, pp. 234–241

    work page 2015

  5. [5]

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

    A. Dosovitskiy et al. , “An image is worth 16x16 words: Transformers for image recognition at scale,” 2020, arXiv:2010.11929

    arXiv 2020

  6. [6]

    Morphological skeleton representation and coding of binary images,

    P. Maragos and R. Schafer, “Morphological skeleton representation and coding of binary images,” IEEE Trans. Acoust. Speech Signal Process. , vol. 34, no. 5, pp. 1228–1244, Oct. 1986

    work page 1986

  7. [7]

    Road extraction from remote sensing imagery based on road tracking and ribbon snake,

    Y . Hu and K.-j. Zu, “Road extraction from remote sensing imagery based on road tracking and ribbon snake,” in Pac.-Asia Conf. Knowl. Eng. Softw. Eng. , Dec. 2009, pp. 201–204

    work page 2009

  8. [8]

    Morphological road segmentation in urban areas from high resolution satellite images,

    R. Gaetano, J. Zerubia, G. Scarpa, and G. Poggi, “Morphological road segmentation in urban areas from high resolution satellite images,” in Int. Conf. Digit. Signal Process. (DSP) , Jul. 2011, pp. 1–8

    work page 2011

Show all 57 references
  1. [9]

    Use of watersheds in contour detection,

    S. Beucher, “Use of watersheds in contour detection,” in Proc. Int. Workshop Image Process. , Sep. 1979, pp. 17–21

    1979

  2. [10]

    clDice-a novel topology-preserving loss function for tubular structure segmentation,

    S. Shit et al. , “clDice-a novel topology-preserving loss function for tubular structure segmentation,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR) , Jun. 2021, pp. 16 560–16 569

    2021

  3. [11]

    Cp-loss: Connectivity-preserving loss for road curb detection in autonomous driving with aerial images,

    Z. Xu, Y . Sun, L. Wang, and M. Liu, “Cp-loss: Connectivity-preserving loss for road curb detection in autonomous driving with aerial images,” in IEEE/RSJ Int. Conf. Intell. Robots Syst. (IROS) , Sep. 2021, pp. 1117– 1123

    2021

  4. [12]

    RoadNet: Learning to comprehensively analyze road networks in complex urban scenes from high-resolution remotely sensed images,

    Y . Liu, J. Yao, X. Lu, M. Xia, X. Wang, and Y . Liu, “RoadNet: Learning to comprehensively analyze road networks in complex urban scenes from high-resolution remotely sensed images,” IEEE Trans. Geosci. Remote Sens., vol. 57, no. 4, pp. 2043–2056, Apr. 2018

    2018

  5. [13]

    Language models are few-shot learners,

    T. Brown et al. , “Language models are few-shot learners,” Adv. Neural Inf. Process. Syst. (NIPS) , vol. 33, pp. 1877–1901, Dec. 2020

    1901

  6. [14]

    Learning transferable visual models from natural language supervision,

    A. Radford et al. , “Learning transferable visual models from natural language supervision,” in Int. Conf. Mach. Learn. , Jul. 2021, pp. 8748– 8763

    2021

  7. [15]

    Segment anything,

    A. Kirillov et al. , “Segment anything,” in Proc. IEEE/CVF Int. Conf. Comput. Vis. (ICCV) , Oct. 2023, pp. 4015–4026

    2023

  8. [16]

    Segment anything, from space?

    S. Ren et al. , “Segment anything, from space?” in Proc. IEEE/CVF Winter Conf. Appl. Comput. Vis. (WACV) , Jan. 2024, pp. 8355–8365

    2024

  9. [17]

    Personalize segment anything model with one shot,

    R. Zhang et al. , “Personalize segment anything model with one shot,” 2023, arXiv:2305.03048

    2023 arXiv

  10. [18]

    The segment anything model (SAM) for remote sensing applications: From zero to one shot,

    L. P. Osco, Q. Wu, E. L. de Lemos, W. N. Gonc ¸alves, A. P. M. Ramos, J. Li, and J. M. Junior, “The segment anything model (SAM) for remote sensing applications: From zero to one shot,” Int. J. Appl. Earth Obs. Geoinf., vol. 124, p. 103540, Nov. 2023

    2023

  11. [19]

    RSPrompter: Learning to prompt for remote sensing instance segmentation based on visual foundation model,

    K. Chen et al. , “RSPrompter: Learning to prompt for remote sensing instance segmentation based on visual foundation model,” IEEE Trans. Geosci. Remote Sens. , vol. 62, pp. 1–17, Jan. 2024

    2024

  12. [20]

    SAM fails to segment anything?–SAM-adapter: Adapt- ing SAM in underperformed scenes: Camouflage, shadow, medical image segmentation, and more,

    T. Chen et al. , “SAM fails to segment anything?–SAM-adapter: Adapt- ing SAM in underperformed scenes: Camouflage, shadow, medical image segmentation, and more,” arXiv:2304.09148, 2023

    2023 arXiv

  13. [21]

    Building extraction from high spatial resolution remote sensing images via multiscale-aware and segmentation-prior conditional random fields,

    Q. Zhu, Z. Li, Y . Zhang, and Q. Guan, “Building extraction from high spatial resolution remote sensing images via multiscale-aware and segmentation-prior conditional random fields,” Remote Sens. , vol. 12, no. 23, p. 3983, Dec. 2020

    2020

  14. [22]

    SAM- Assisted remote sensing imagery semantic segmentation with object and boundary constraints,

    X. Ma, Q. Wu, X. Zhao, X. Zhang, M.-O. Pun, and B. Huang, “SAM- Assisted remote sensing imagery semantic segmentation with object and boundary constraints,” IEEE Trans. Geosci. Remote Sens. , vol. 62, pp. 1–16, Aug. 2024

    2024

  15. [23]

    Algorithm unrolling: Interpretable, efficient deep learning for signal and image processing,

    V . Monga, Y . Li, and Y . C. Eldar, “Algorithm unrolling: Interpretable, efficient deep learning for signal and image processing,” IEEE Signal Process. Mag., vol. 38, no. 2, pp. 18–44, Feb. 2021

    2021

  16. [24]

    Conditional random fields as recurrent neural net- works,

    S. Zheng et al. , “Conditional random fields as recurrent neural net- works,” in Proc. IEEE Int. Conf. Comput. Vis. , Dec. 2015, pp. 1529– 1537

    2015

  17. [25]

    A regularized convolutional neural network for semantic image segmentation,

    F. Jia, J. Liu, and X.-C. Tai, “A regularized convolutional neural network for semantic image segmentation,” Anal. Appl., vol. 19, no. 01, pp. 147– 165, 2021

    2021

  18. [26]

    Interpretable small training set image segmentation network originated from multi-grid variational model,

    J. Meng, W. Guo, J. Liu, and M. Yang, “Interpretable small training set image segmentation network originated from multi-grid variational model,” 2023, arXiv:2306.14097

    2023 arXiv

  19. [27]

    Deep convolutional neural networks with spatial regularization, volume and star-shape priors for image segmentation,

    J. Liu, X. Wang, and X.-C. Tai, “Deep convolutional neural networks with spatial regularization, volume and star-shape priors for image segmentation,” J. Math. Imaging Vis. , vol. 64, no. 6, pp. 625–645, Apr. 2022

    2022

  20. [28]

    Boundary- enhanced dual-stream network for semantic segmentation of high- resolution remote sensing images,

    X. Li, L. Xie, C. Wang, J. Miao, H. Shen, and L. Zhang, “Boundary- enhanced dual-stream network for semantic segmentation of high- resolution remote sensing images,” GISci. Remote Sens. , vol. 61, no. 1, p. 2356355, Jul. 2024

    2024

  21. [29]

    Topology-guided road graph extraction from remote sensing images,

    Y . Zao, Z. Zou, and Z. Shi, “Topology-guided road graph extraction from remote sensing images,” IEEE Trans. Geosci. Remote Sens. , vol. 62, pp. 1–14, Dec. 2023

    2023

  22. [30]

    Semantic segmentation of remote sensing images by interactive representation refinement and geometric prior-guided inference,

    X. Li, F. Xu, F. Liu, Y . Tong, X. Lyu, and J. Zhou, “Semantic segmentation of remote sensing images by interactive representation refinement and geometric prior-guided inference,” IEEE Trans. Geosci. Remote Sens. , vol. 62, pp. 1–18, Dec. 2023

    2023

  23. [31]

    Swint-resnet: An improved remote sensing image segmentation model based on transformer,

    Y . Ma, Y . Wang, X. Liu, and H. Wang, “Swint-resnet: An improved remote sensing image segmentation model based on transformer,” IEEE Geosci. Remote Sens. Lett. , vol. 21, pp. 1–5, Jul. 2024

    2024

  24. [32]

    Nonlinear total variation based noise removal algorithms,

    L. I. Rudin, S. Osher, and E. Fatemi, “Nonlinear total variation based noise removal algorithms,” Physica D , vol. 60, no. 1-4, pp. 259–268, Nov. 1992

    1992

  25. [33]

    Active contours without edges,

    T. F. Chan and L. A. Vese, “Active contours without edges,” IEEE Trans. Image Process., vol. 10, no. 2, pp. 266–277, Feb. 2001

    2001

  26. [34]

    Fast global minimization of the active contour/snake model,

    X. Bresson, S. Esedo ¯glu, P. Vandergheynst, J.-P. Thiran, and S. Osher, “Fast global minimization of the active contour/snake model,” J. Math. Imaging Vis., vol. 28, no. 2, pp. 151–167, Jul. 2007

    2007

  27. [35]

    Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images,

    S. Geman and D. Geman, “Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 6, no. 6, pp. 721–741, Nov. 1984

    1984

  28. [36]

    Spatial models for fuzzy clustering,

    D. L. Pham, “Spatial models for fuzzy clustering,” Comput. Vis. Image Underst., vol. 84, no. 2, pp. 285–297, Nov. 2001

    2001

  29. [37]

    An edge-weighted centroidal voronoi tessellation model for image segmentation,

    L. J. J. Wang and X. Wang, “An edge-weighted centroidal voronoi tessellation model for image segmentation,”IEEE Trans. Image Process., vol. 18, no. 8, pp. 1844–1858, Aug. 2009

    2009

  30. [38]

    A fast segmentation method based on constraint optimization and its applications: Intensity inhomogeneity and texture segmentation,

    H. H. J. Liu, X.-C. Tai and Z. Huan, “A fast segmentation method based on constraint optimization and its applications: Intensity inhomogeneity and texture segmentation,” Pattern Recognit., vol. 44, no. 9, pp. 2093– 2108, Sep. 2011

    2011

  31. [39]

    Image segmentation using a local GMM in a variational framework,

    J. Liu and H. Zhang, “Image segmentation using a local GMM in a variational framework,” J. Math. Imaging Vis. , vol. 46, no. 2, pp. 161– 176, Aug. 2012

    2012

  32. [40]

    Merriman, J

    B. Merriman, J. K. Bence, and S. Osher, Diffusion generated motion by mean curvature. Department of Mathematics, University of California, Los Angeles, 1992. 13

    1992

  33. [41]

    Threshold dynamics for the piecewise constant Mumford-Shah functional,

    S. Esedog Lu and Y .-H. R. Tsai, “Threshold dynamics for the piecewise constant Mumford-Shah functional,” J. Comput. Phys. , vol. 211, no. 1, pp. 367–384, Jan. 2006

    2006

  34. [42]

    An efficient iterative thresholding method for image segmentation,

    D. Wang, H. Li, X. Wei, and X.-P. Wang, “An efficient iterative thresholding method for image segmentation,” J. Comput. Phys. , vol. 350, pp. 657–667, Dec. 2017

    2017

  35. [43]

    Image analysis using mathematical morphology,

    R. M. Haralick, S. R. Sternberg, and X. Zhuang, “Image analysis using mathematical morphology,” IEEE Trans. Pattern Anal. Mach. Intell. , no. 4, pp. 532–550, Jul. 1987

    1987

  36. [44]

    R. C. Gonzales and P. Wintz, Digital image processing . Addison- Wesley Longman Publishing Co., Inc., 1987

    1987

  37. [45]

    Masked autoencoders are scalable vision learners,

    K. He, X. Chen, S. Xie, Y . Li, P. Doll ´ar, and R. Girshick, “Masked autoencoders are scalable vision learners,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR) , Jun. 2022, pp. 16 000–16 009

    2022

  38. [46]

    RingMo-SAM: A foundation model for segment anything in multimodal remote-sensing images,

    Z. Yan et al., “RingMo-SAM: A foundation model for segment anything in multimodal remote-sensing images,” IEEE Trans. Geosci. Remote Sens., vol. 61, pp. 1–16, Nov. 2023

    2023

  39. [47]

    Lora: Low-rank adaptation of large language models,

    E. J. Hu et al. , “Lora: Low-rank adaptation of large language models,” 2021, arXiv:2106.09685

    2021 arXiv

  40. [48]

    Customized segment anything model for medical image segmentation,

    K. Zhang and D. Liu, “Customized segment anything model for medical image segmentation,” 2023, arXiv:2304.13785

    2023 arXiv

  41. [49]

    On the numerical solution of heat conduction problems in two and three space variables,

    J. Douglas and H. H. Rachford, “On the numerical solution of heat conduction problems in two and three space variables,” Trans. Amer . Math. Soc. , vol. 82, no. 2, pp. 421–439, Jul. 1956

    1956

  42. [50]

    Augmented Lagrangian method, dual methods, and split Bregman iteration for ROF, vectorial TV, and high order models,

    C. Wu and X.-C. Tai, “Augmented Lagrangian method, dual methods, and split Bregman iteration for ROF, vectorial TV, and high order models,” SIAM J. Imaging Sci. , vol. 3, no. 3, pp. 300–339, Jul. 2010

    2010

  43. [51]

    Glowinski, T.-W

    R. Glowinski, T.-W. Pan, and X.-C. Tai, Some Facts About Operator- Splitting and Alternating Direction Methods . Springer, Jan. 2016, pp. 19–94

    2016

  44. [52]

    Glowinski and P

    R. Glowinski and P. Le Tallec, Augmented Lagrangian and operator- splitting methods in nonlinear mechanics . SIAM, 1989

    1989

  45. [53]

    Fully convolutional networks for multisource building extraction from an open aerial and satellite imagery data set,

    S. Ji, S. Wei, and M. Lu, “Fully convolutional networks for multisource building extraction from an open aerial and satellite imagery data set,” IEEE Trans. Geosci. Remote Sens. , vol. 57, no. 1, pp. 574–586, Aug. 2018

    2018

  46. [54]

    Learning aerial image segmentation from online maps,

    P. Kaiser, J. D. Wegner, A. Lucchi, M. Jaggi, T. Hofmann, and K. Schindler, “Learning aerial image segmentation from online maps,” IEEE Trans. Geosci. Remote Sens. , vol. 55, no. 11, pp. 6054–6068, Jul. 2017

    2017

  47. [55]

    Machine learning for aerial image labeling,

    V . Mnih, “Machine learning for aerial image labeling,” Ph.D. disserta- tion, University of Toronto, CAN, 2013, aAINR96184

    2013

  48. [56]

    MCANet: A joint semantic segmentation framework of optical and SAR images for land use classification,

    X. Li et al. , “MCANet: A joint semantic segmentation framework of optical and SAR images for land use classification,” Int. J. Appl. Earth Obs. Geoinf. , vol. 106, p. 102638, Feb. 2022. Jun Xie received the B.S. degree in mathematics from the School of Mathematical Sciences a...

    2022

  49. [2004]

    and Ph.D

    He received the M.S. and Ph.D. degrees in computational mathematics from the Beijing Normal University (BNU), China, in 2008 and 2011 respec- tively. He is currently an associate professor at BNU. His research interests include variational, optimal transport and deep learning ...

    2008

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.