Pith.

Computer Vision and Pattern Recognition · arXiv:2611.17077 v1 · submitted 12 Nov 2026 · 38 references

SALT: State-Aligned Low-Rank Transfer for Single-Temporal Cross-Dataset Structural Localization in Remote Sensing

Aggelos Psiris, Georgios Angelidis, Giorgos Sfikas, Maria Vakalopoulou, Thomas Lagkas, Panagiotis Sarigiannidis, Georgios Th. Papadopoulos

The paper says

“Change detection in remote sensing conventionally relies on spatially aligned bi-temporal imagery, which may be unavailable or unreliable in practical scenarios.”

more of the abstract

Recent single-temporal approaches address this limitation by learning structural evidence, such as building regions, from a labeled source dataset and transferring it to unseen target imagery using only the current image. However, adaptation to the source dataset can over-specialize pretrained features and reduce their cross-dataset generalization, a risk that this paper argues is higher when the feature adaptation is parameterized independently of the states used for prediction. To address this issue, the current study introduces SALT, a State-Aligned Low-Rank Transfer framework. SALT keeps a DINOv3 backbone frozen and learns negative and positive structural states that both score each patch and, through State-Aligned Low-Rank Adaptation (SALA), span the rank-2 subspace in which patch features are corrected. Directional State Alignment supervises these corrections to raise the evidence of positive patches and lower that of negative ones, while Geometry Preservation limits changes to the pairwise relations among pretrained patch features. Experiments on LEVIR-CD and the WHU aerial building dataset show that, at the pixel level and in both transfer directions, SALT outperforms four baselines reproduced under a common source-only protocol, improving over the strongest reproduced baselines by 9.45 and 5.79 points in pixel-level AP and by 27.69 and 14.25 points in IoU (LEVIR-CD to WHU and WHU to LEVIR-CD, respectively), with only 0.068M trainable parameters and 40.2 ms per image. In order to facilitate reproducibility and further research, the source code of SALT is available at https://github.com/gthpapadopoulos/SALT

from the abstract

Pith says

Machine review · deepseek-v4-flash · 13 Nov 2026 · what was read was not recorded · not checked by a person

It can be wrong. Challenge this review or reply to it.

Holds up, with conditions.

3 major objections · 5 minor

How often objections from reviews like this one are right: not yet measured

Are you an author? What you can do.

Not checked: figures, code, what the 38 references say.

The paper says, at §4.2, Table 1

The review did not quote this passage.

Pith says: Objection 1 of 3 · major

The central comparative claim rests on an unmatched tuning budget. SALT's hyperparameters are selected per transfer direction on source validation via a staged procedure (Section S2), while baselines use author-recommended defaults with no comparable validation tuning. Because the margins (9.45 and 5.79 P-AP points) are the headline numbers, this asymmetry is load-bearing.

more of objection 1

The paper does not acknowledge this limitation. A fairness check—e.g., tuning baselines on source validation or evaluating SALT with default hyperparameters—is needed to assess whether the reported advantage is robust or partly an artifact of tuning effort.

The paper says, at §3.4, Eq. (3)-(4)

The review did not quote this passage.

A classifier read this sentence of the paper as acknowledging the concern (probability 0.80):

“Limitations and Future Work The current formulation represents structural evidence with a single pair of negative and positive states, which restricts SALA to a rank-2 adaptation space and may not capture heterogeneous positive appearances.”

Pith says: Objection 2 of 3 · major

The rank-2 adaptation space is constructed from two learned states via Gram-Schmidt. The paper acknowledges in §5 that this 'may not capture heterogeneous positive appearances.' This is a genuine limitation, but the paper does not quantify how often the required adaptation lies outside this subspace.

more of objection 2

The ablation in Table S9 shows that State rank-1 (single direction) is competitive with SALT, suggesting the second direction adds modest benefit. The authors should discuss under what conditions the rank-2 assumption might fail and whether the method could be extended to adaptive rank.

The paper says, at §4.2, Table 1

The review did not quote this passage.

A classifier read this sentence of the paper as acknowledging the concern (probability 0.60):

“All model parameters remain fixed, and binary predictions use a threshold selected on source validation data.”

Pith says: Objection 3 of 3 · major

The F1 and IoU comparisons use thresholds selected on source validation, whereas the baselines FIFA-RS and BSSRS originally used target-optimal thresholds. The paper correctly notes this difference, but the reported F1/IoU margins (27.69 and 14.25 points) are inflated relative to the original evaluation protocol. The authors should report both source-selected and target-optimal thresholds for all methods to enable fair comparison with prior work.

5 minor objections

The paper says, at §1, Fig. 1

Pith says: minor objection 1 of 5

Figure 1 uses a logarithmic scale for trainable parameters, which visually exaggerates the difference between SALT and baselines. Consider using a linear scale or a table for precise comparison.

The paper says, at §3.5, Eq. (6)

Pith says: minor objection 2 of 5

The notation [x]+ is defined as max(0, x), but the equation uses [m - Δe_i]+ and [m + Δe_i]+. Clarify that the margin m is positive and the loss encourages Δe_i ≥ m for positive patches and Δe_i ≤ -m for negative patches.

show 3 more

The paper says, at §4.2, Table 2

Pith says: minor objection 3 of 5

The training time for WHU→LEVIR-CD is reported in hours, while LEVIR-CD→WHU is in minutes. This is due to dataset size differences, but it would be helpful to note this explicitly in the caption.

The paper says, at §5

Pith says: minor objection 4 of 5

The limitations section is well-written and honest. However, the paper could also mention that the evaluation is limited to building-related structures (changed buildings vs. building footprints) and may not generalize to other structural types (e.g., roads, water bodies).

The paper says, at Supplementary S2, Table S8

Pith says: minor objection 5 of 5

The final configurations differ substantially between the two source datasets (e.g., α=0.25 vs 2.00, m=0.10 vs 0.20, λ_dir=0.50 vs 1.00). This suggests the method is sensitive to hyperparameters. The paper should discuss whether a single configuration could work for both directions or whether per-dataset tuning is always required.

What would settle it

Retune one same-backbone baseline (VisualAD or AVA-DINO, both on frozen DINOv3 ViT-L/16) on each source validation split using the identical staged procedure of Section S2 (hyperparameters selected per direction by P-AP), then evaluate on the target under the same three seeds and source-selected thresholds. If the retuned baseline's target P-AP rises by more than about two points in either direction, the 9.45/5.79-point margins in Table 1 are partly an artifact of unmatched tuning and the comparative claim must be restated with equalized tuning budgets. A complementary check: run SALT on each direction with the other direction's tuned configuration (e.g., LEVIR→WHU with α=2.0, m=0.2); the resulting drop quantifies how much of SALT's margin is attributable to tuning rather than to the SALA coupling itself.

from the skeptic's reading

How the review went

3 readings by deepseek-v4-flash · 13 Nov 2026

  1. Reader holds up, with conditions confidence high
  2. Skeptic partly agrees 53 s · 19,772 tokens read, 8,843 written
  3. Referee major revision: 3 major objections, 5 minor 13 s · 20,143 tokens read, 1,086 written

Not run on this paper: the editor, the simulated authors' defence.

Pith did not check this review's objections against the paper.

The review in full

The strongest objection

The headline claim — that SALT outperforms four reproduced baselines under a common source-only protocol — rests on a comparison whose tuning budgets are not matched. SALT's configuration is selected per transfer direction on the source validation split through the staged, P-AP-guided procedure of Section S2 (backbone, λdir, m, λgeo, α, ρ, σg); the settings differ sharply between directions (α = 0.25 vs 2.00, m = 0.10 vs 0.20, λdir = 0.50 vs 1.00), so tuning demonstrably moves results. The four baselines are evaluated with author-recommended defaults and no comparable validation tuning. Because method quality is confounded with tuning budget, the threshold-free core of the claim — the P-AP margins of 9.45 and 5.79 points in Table 1 — may overstate SALT's genuine advantage; they could shrink materially if baselines received the same validation-based tuning. This asymmetry is not acknowledged anywhere in the paper. It is distinct from the threshold-protocol caveat, which affects only the F1/IoU comparisons, so it targets exactly the numbers the headline rests on.

I read the paper in good faith. The central claim is a performance claim: SALT reports the highest pixel-level localization among four reproduced baselines in both transfer directions at 0.068M trainable parameters. The architecture is clearly described, the frozen-backbone design and released code give independent support, and key controls are already in place: VisualAD and AVA-DINO share SALT's DINOv3 backbone, and the P-AP superiority is threshold-free, so the threshold-protocol caveat (F1/IoU only) does not threaten the headline. The reader's flagged weak assumption — that the rank-2 state-aligned space suffices — is acknowledged by the authors (Section 5) and is partly addressed by their own data: Table S9 shows state-aligned rank-1 matches or beats SALT on several metrics, and Table S10 shows the two states are nearly anti-parallel (cos ≈ −0.89/−0.92), so the second Gram-Schmidt direction has norm ≈ 0.2 and the construction is nearly rank-1 in practice. That weakens the explanatory story but not the headline performance, since even the single state-aligned direction wins. I therefore do not treat rank-2 sufficiency as the most load-bearing issue for the strongest claim. The most load-bearing issue is comparator fairness: SALT's per-direction, P-AP-guided validation tuning (S2) is not matched by the baselines, which run with author defaults. Since the tuned settings differ 8× in α and 2× in margin between directions, tuning is doing real work, and the untuned baselines may be handicapped. This directly supports the size of the claimed margins, is not acknowledged, and is testable by retuning one same-backbone baseline. If the margin survives equalized tuning, the claim stands; if not, the verdict should carry the condition that the comparison be re-run with matched tuning budgets. Either way the verdict stays CONDITIONAL — the method is sound and well-documented, but the evidence base is one dataset pair and the comparative margins need this fairness check.

The referee's report

If the results hold, SALT offers a parameter-efficient approach to cross-dataset transfer for structural localization, which is practically relevant for remote sensing applications where bi-temporal imagery may be unavailable. The idea of coupling feature adaptation with the same states used for prediction is novel and could inspire further work. The paper provides thorough ablation studies and analysis.

Pith's reading of the paper (a machine's summary, not the authors')

The paper introduces SALT, a framework for single-temporal cross-dataset structural localization in remote sensing. It keeps a frozen DINOv3 backbone and learns two structural states (negative and positive) that serve dual roles: they score each image patch and define the rank-2 subspace in which patch features are corrected. This coupling between prediction and adaptation is the core idea. Experiments on LEVIR-CD and WHU datasets show that SALT outperforms four reproduced baselines in both transfer directions, improving pixel-level AP by 9.45 and 5.79 points and IoU by 27.69 and 14.25 points, with only 0.068M trainable parameters. The paper argues that state-aligned adaptation prevents over-specialization and preserves pretrained feature relationships better than unrestricted or generic low-rank adaptation.

What carries the argument. State-Aligned Low-Rank Adaptation (SALA): a mechanism that constructs a rank-2 adaptation subspace from the Gram-Schmidt orthogonalization of the two learned structural states (negative and positive). Each patch feature is corrected by a linear combination of these basis vectors, with coefficients predicted by a lightweight MLP. This couples feature adaptation to the prediction states, limiting changes to directions relevant to the structural task.

Core claim. The central claim is that feature adaptation for cross-dataset transfer should be constrained by the same structural states used for prediction. SALT introduces State-Aligned Low-Rank Adaptation (SALA), where the learned negative and positive structural states are orthogonalized to form a rank-2 basis. Each patch feature is corrected by a linear combination of these basis vectors, with coefficients predicted by a lightweight MLP. Directional State Alignment (DSA) supervises these corrections to increase positive evidence and decrease negative evidence, while Geometry Preservation maintains pairwise similarities among pretrained features. The paper demonstrates that this state-aligned approach outperforms unrestricted adaptation (Matched MLP), independently learned low-rank bases (Free rank-2), and even a single state-derived direction (State rank-1) on both transfer directions.

The premise it rests on. The rank-2 subspace spanned by the two learned structural states is sufficient to capture the relevant feature corrections needed for cross-dataset transfer, meaning that any domain shift requiring feature adjustments outside this two-dimensional space would not be expressible by the method.

If the paper is right (Pith's inferences, not the authors')

SALT provides a new paradigm for parameter-efficient transfer learning in remote sensing, where the adaptation space is explicitly tied to the task's prediction states, potentially applicable to other structural localization tasks beyond building change detection.

The method's extreme parameter efficiency (0.068M trainable parameters) and low latency (40.2 ms per image) make it suitable for deployment on resource-constrained platforms like edge devices or satellites.

State-aligned adaptation could be extended to other domains where pretrained features need task-specific correction while preserving generalization, such as medical image analysis or industrial anomaly detection.

The paper suggests that the rank-2 constraint is sufficient for the tested datasets, implying that many cross-dataset shifts in remote sensing may be captured by a low-dimensional correction subspace.

The success of rank-2 adaptation suggests that the domain shift between LEVIR-CD and WHU is largely captured by a two-dimensional correction in feature space, which may be a general property of remote sensing datasets with similar semantic structures.

SALT's reliance on a single pair of states may limit its ability to handle heterogeneous positive appearances (acknowledged by the authors). An extension with multiple context-dependent states or adaptive-rank adaptation could improve performance on more diverse target domains.

The single-temporal setting cannot distinguish true temporal change from source-defined structural evidence, which likely explains the lower performance when transferring from WHU (building footprints) to LEVIR-CD (building changes). This limitation is inherent to the problem formulation, not just the method.

Check this review

Pith stored no record of this review's passes and took no fingerprint of it. Reviews made by code from 13 Nov 2026 on carry both.

How often reviews like this one are right

Reviews like this one: written by deepseek-v4-flash, where what was read was not recorded.

How often objections from reviews like this one are right: not yet measured

Every value above, in the review's public record (JSON)

Reviews by people

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review

How Pith checks papers · The full audit of this paper

The paper

Abstract

Change detection in remote sensing conventionally relies on spatially aligned bi-temporal imagery, which may be unavailable or unreliable in practical scenarios. Recent single-temporal approaches address this limitation by learning structural evidence, such as building regions, from a labeled source dataset and transferring it to unseen target imagery using only the current image. However, adaptation to the source dataset can over-specialize pretrained features and reduce their cross-dataset generalization, a risk that this paper argues is higher when the feature adaptation is parameterized independently of the states used for prediction. To address this issue, the current study introduces SALT, a State-Aligned Low-Rank Transfer framework. SALT keeps a DINOv3 backbone frozen and learns negative and positive structural states that both score each patch and, through State-Aligned Low-Rank Adaptation (SALA), span the rank-2 subspace in which patch features are corrected. Directional State Alignment supervises these corrections to raise the evidence of positive patches and lower that of negative ones, while Geometry Preservation limits changes to the pairwise relations among pretrained patch features. Experiments on LEVIR-CD and the WHU aerial building dataset show that, at the pixel level and in both transfer directions, SALT outperforms four baselines reproduced under a common source-only protocol, improving over the strongest reproduced baselines by 9.45 and 5.79 points in pixel-level AP and by 27.69 and 14.25 points in IoU (LEVIR-CD to WHU and WHU to LEVIR-CD, respectively), with only 0.068M trainable parameters and 40.2 ms per image. In order to facilitate reproducibility and further research, the source code of SALT is available at https://github.com/gthpapadopoulos/SALT

The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.

abstract compiled · 2026-11-14

The authors' main claim, as the classifier picked it out 1.00

“Experiments on LEVIR-CD and the WHU aerial building dataset show that, at the pixel level and in both transfer directions, SALT outperforms four baselines reproduced under a common source-only protocol, improving over the strongest reproduced baselines by 9.45 and 5.79 points in pixel-level AP and by 27.69 and 14.25 points in IoU (LEVIR-CD to WHU and WHU to LEVIR-CD, respectively), with only 0.068M trainable parameters and 40.2 ms per image.”

How the main claim is stated: asserted 1.00

What the abstract itself states, as probabilities: claims first 0.05 · states limitation 0.12 · human participants 0.02 · controlled comparison 0.95 · randomized assignment 0.01 · states quantitative result 0.99 · compares with prior results 0.98 · mentions code or data availability 0.99

The abstract, sentence by sentence

This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.

SentenceRoleThe authors' own workStates the main conclusionStrength
Change detection in remote sensing conventionally relies on spatially aligned bi-temporal imagery, which may be unavailable or unreliable in practical scenarios. background 0.99 0.04 0.02 hedged possibility 0.86
Recent single-temporal approaches address this limitation by learning structural evidence, such as building regions, from a labeled source dataset and transferring it to unseen target imagery using only the current image. background 1.00 0.05 0.03 asserted 0.99
However, adaptation to the source dataset can over-specialize pretrained features and reduce their cross-dataset generalization, a risk that this paper argues is higher when the feature adaptation is parameterized independently of the states used for prediction. gap 0.77 0.14 0.16 hedged possibility 0.68
To address this issue, the current study introduces SALT, a State-Aligned Low-Rank Transfer framework. contribution 0.99 0.85 0.07 asserted 0.99
SALT keeps a DINOv3 backbone frozen and learns negative and positive structural states that both score each patch and, through State-Aligned Low-Rank Adaptation (SALA), span the rank-2 subspace in which patch features are corrected. method 0.97 0.91 0.06 asserted 0.99
Directional State Alignment supervises these corrections to raise the evidence of positive patches and lower that of negative ones, while Geometry Preservation limits changes to the pairwise relations among pretrained patch features. method 1.00 0.81 0.06 asserted 0.98
Experiments on LEVIR-CD and the WHU aerial building dataset show that, at the pixel level and in both transfer directions, SALT outperforms four baselines reproduced under a common source-only protocol, improving over the strongest reproduced baselines by 9.45 and 5.79 points in pixel-level AP and by 27.69 and 14.25 points in IoU (LEVIR-CD to WHU and WHU to LEVIR-CD, respectively), with only 0.068M trainable parameters and 40.2 ms per image. result 1.00 0.97 0.74 asserted 1.00
In order to facilitate reproducibility and further research, the source code of SALT is available at https://github.com/gthpapadopoulos/SALT future work 0.82 0.70 0.02 asserted 0.99

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.97

What was examined, and what was not

Sources read

  • arXiv listing abstract v1

Not examined

  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers

Levels run

  • P abstract

Not run

  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

How the labels were made

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-11-14: 1 request, 6,599 input tokens.

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON

Figures (3)
Figure 1 of the paper
Figure 1. Localization performance versus adaptation size. Pixel-level AP versus trainable parameters (logarithmic scale). Colors identify the methods; circles and triangles denote LEVIR￾CD→WHU and WHU→LEVIR-CD, respectively. Thin vertical lines connect each method’s two transfer results, and error bars show standard deviations over three seeds. SALT achieves the highest pixel AP in both directions with 69× to 465× fewer train￾able parameters than the reproduced baselines. further research, the source code of SALT is available at https://github.com/gthpapadopoulos/SALT 1. Introduction Change d
Figure 2 of the paper
Figure 2. Overview of SALT. A frozen DINOv3 extracts patch features from a single image. SALA uses the learned negative and positive structural states to define a rank-2 adaptation space, within which each patch feature is corrected. The adapted features are then scored to produce a dense structural response map. The four training objectives (top) supervise structural predictions, guide feature corrections, preserve feature relationships, and separate the learned states. Green arrows indicate training on the source dataset, while red arrows indicate inference on the unseen target dataset.…
Figure 3 of the paper
Figure 3. Indicative cross-dataset localization results. Rows are labeled by the target dataset (WHU rows from LEVIR-CD

Figures are the authors', reproduced from the paper.

References (38)
  1. [1] Muhammad Aqeel, Maham Nazir, Uzair Khan, Marco Cristani, and Francesco Setti. Anomaly-aware vision- language adapters for zero-shot anomaly detection. In2026 IEEE International Conference on Image Processing (ICIP), pages 1–6, 2026. 2, 3, 6, 7 on Pith
  2. [2] Yunkang Cao, Jiangning Zhang, Luca Frittoli, Yuqi Cheng, Weiming Shen, and Giacomo Boracchi. AdaCLIP: Adapting CLIP with hybrid learnable prompts for zero-shot anomaly detection. InEuropean Conference on Computer Vision, pages 55–72. Springer, 2024. 3 on Pith
  3. [3] Hao Chen and Zhenwei Shi. A spatial-temporal attention- based method and a new dataset for remote sensing image change detection.Remote Sensing, 12(10):1662, 2020. 1, 2, 6 on Pith
  4. [4] Rodrigo Caye Daudt, Bertrand Le Saux, and Alexandre Boulch. Fully convolutional siamese networks for change detection. In2018 25th IEEE international conference on image processing (ICIP), pages 4063–4067. IEEE, 2018. 2 on Pith
  5. [5] Sijun Dong, Libo Wang, Bo Du, and Xiaoliang Meng. ChangeCLIP: Remote sensing change detection with multi- modal vision-language representation learning.ISPRS Jour- nal of Photogrammetry and Remote Sensing, 208:53–69, on Pith
  6. [6] Qiangang Du, Jinlong Peng, Xu Chen, Qingdong He, Liren He, Qiang Nie, and Mingmin Chi. SCL: Towards domain generalization via single-temporal multimodal contrastive learning for remote sensing change detection. InProceed- ings of the IEEE/CVF Conference on Computer Vision and ... on Pith
  7. [7] Matic Fu ˇcka, Vitjan Zavrtanik, and Danijel Skoˇcaj. Anoma- lyVFM – transforming vision foundation models into zero- shot anomaly detectors. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 35555–35566, 2026. 3 on Pith
  8. [8] Yanning Hou, Peiyuan Li, Zirui Liu, Yitong Wang, Yanran Ruan, Jianfeng Qiu, and Ke Xu. VisualAD: Language-free zero-shot anomaly detection via vision transformer. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 21346–21356, 2... on Pith
  9. [9] Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 8 LoRA: Low-rank adaptation of large language models. InIn- ternational Conference on Learning Representations, 2022. 3, 4 on Pith
  10. [10] Jongheon Jeong, Yang Zou, Taewan Kim, Dongqing Zhang, Avinash Ravichandran, and Onkar Dabeer. WinCLIP: Zero- /few-shot anomaly classification and segmentation. In2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 19606–19616. IEEE, 2023. 3 on Pith
  11. [11] Shunping Ji, Shiqing Wei, and Meng Lu. Fully convolutional networks for multisource building extraction from an open aerial and satellite imagery data set.IEEE Transactions on Geoscience and Remote Sensing, 57(1):574–586, 2019. 2, 6, 8 on Pith
  12. [12] Ananya Kumar, Aditi Raghunathan, Robbie Matthew Jones, Tengyu Ma, and Percy Liang. Fine-tuning can distort pre- trained features and underperform out-of-distribution. InIn- ternational Conference on Learning Representations, 2022. 3 on Pith
  13. [13] Jingtao Li, Xinyu Wang, Hengwei Zhao, and Yanfei Zhong. Learning a cross-modality anomaly detector for remote sens- ing imagery.IEEE Transactions on Image Processing, 33: 6607–6621, 2024. 3 on Pith
  14. [14] Kaiyu Li, Xiangyong Cao, and Deyu Meng. A new learn- ing paradigm for foundation model-based remote-sensing change detection.IEEE Transactions on Geoscience and Re- mote Sensing, 62:1–12, 2024. 2 on Pith
  15. [15] Shih-Chih Lin, Jia-Xian Jian, YunTung Chu, and Wei-Chieh Sun. FIFA-RS: Fine-grained image-feature alignment for structural anomaly reasoning in remote sensing. InProceed- ings of the 39th Canadian Conference on Artificial Intelli- gence, pages 552–563. PMLR, 2026. 2, 3, 6, 7 on Pith
  16. [16] Shih-Chih Lin, Jia-Xian Jian, YunTung Chu, Wei-Chieh Sun, and Fang-Yi Lin. Bridging semantic and structural man- ifolds: Zero-shot single-temporal anomaly detection in re- mote sensing. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) ... on Pith
  17. [17] Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Doll´ar. Focal loss for dense object detection. InPro- ceedings of the IEEE International Conference on Computer Vision, pages 2980–2988, 2017. 5 on Pith
  18. [18] Wenxin Ma, Xu Zhang, Qingsong Yao, Fenghe Tang, Chenxu Wu, Yingtai Li, Rui Yan, Zihang Jiang, and S Kevin Zhou. AA-CLIP: Enhancing zero-shot anomaly detection via anomaly-aware CLIP. In2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 4744–4754.... on Pith
  19. [19] Seunghyun Oh, Seongsu Lee, Seunghye Chae, and Young- min Ro. AULoRA: Anomaly understanding with low-rank adaptation for zero-shot anomaly detection.IEEE Access, 13:170095–170105, 2025. 3 on Pith
  20. [20] Guansong Pang, Chunhua Shen, and Anton van den Hengel. Deep anomaly detection with deviation networks. InPro- ceedings of the 25th ACM SIGKDD International Confer- ence on Knowledge Discovery & Data Mining, pages 353– 362, 2019. 5 on Pith
show 18 more
  1. [21] Jun Yeong Park, JunYoung Seo, Minji Kang, and Yu Rang Park. MoECLIP: Patch-specialized experts for zero-shot anomaly detection. InProceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR), pages 35534–35544, 2026. 2, 3, 6, 7 on Pith
  2. [22] Wonpyo Park, Dongju Kim, Yan Lu, and Minsu Cho. Relational knowledge distillation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3967–3976, 2019. 5 on Pith
  3. [23] Daifeng Peng, Xuelian Liu, Yongjun Zhang, Haiyan Guan, Yansheng Li, and Lorenzo Bruzzone. Deep learning change detection techniques for optical remote sensing imagery: Status, perspectives and challenges.International Jour- nal of Applied Earth Observation and Geoinformation, ... on Pith
  4. [24] Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. InInternational conference on machine learning, p... on Pith
  5. [25] Minseok Seo, Hakjin Lee, Yongjin Jeon, and Junghoon Seo. Self-Pair: Synthesizing changes from single source for ob- ject change detection in remote sensing imagery. InProceed- ings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pages 6374–6383, 2023. 2 on Pith
  6. [26] Oriane Sim ´eoni, Huy V V o, Maximilian Seitzer, Federico Baldassarre, Maxime Oquab, Cijo Jose, Vasil Khalidov, Marc Szafraniec, Seungeun Yi, Micha ¨el Ramamonjisoa, et al. DINOv3.arXiv preprint arXiv:2508.10104, 2025. 2, 3 on PithPith's reviewarXiv
  7. [27] Carole H Sudre, Wenqi Li, Tom Vercauteren, Sebastien Ourselin, and M Jorge Cardoso. Generalised Dice overlap as a deep learning loss function for highly unbalanced seg- mentations. InInternational Workshop on Deep Learning in Medical Image Analysis, pages 240–248. Springer, 2017. 5 on Pith
  8. [28] Frederick Tung and Greg Mori. Similarity-preserving knowl- edge distillation. InProceedings of the IEEE/CVF Interna- tional Conference on Computer Vision (ICCV), pages 1365– 1374, 2019. 5 on Pith
  9. [29] Zhixiang Wei, Lin Chen, Yi Jin, Xiaoxiao Ma, Tianle Liu, Pengyang Ling, Ben Wang, Huaian Chen, and Jinjin Zheng. Stronger, fewer, & superior: Harnessing vision foundation models for domain generalized semantic segmentation. In Proceedings of the IEEE/CVF Conference on Computer... on Pith
  10. [30] Mitchell Wortsman, Gabriel Ilharco, Jong Wook Kim, Mike Li, Simon Kornblith, Rebecca Roelofs, Raphael Gon- tijo Lopes, Hannaneh Hajishirzi, Ali Farhadi, Hongseok Namkoong, and Ludwig Schmidt. Robust fine-tuning of zero-shot models. InProceedings of the IEEE/CVF Confer- ence on... on Pith
  11. [31] Zhengxuan Wu, Aryaman Arora, Zheng Wang, Atticus Geiger, Dan Jurafsky, Christopher D Manning, and Christo- pher Potts. ReFT: Representation finetuning for language models.Advances in Neural Information Processing Sys- tems, 37:63908–63962, 2024. 3 on Pith
  12. [32] Jianming Xu, Kai Yan, Zaiwang Fan, Kun Jia, Jianbo Qi, Biao Cao, Wenzhi Zhao, Guoqiang Wang, and Qiao Wang. 9 Toward a novel method for general on-orbit Earth sur- face anomaly detection leveraging large vision models and lightweight priors.IEEE Transactions on Geoscience and ... on Pith
  13. [33] Yu Zhang and Zhi Gao. RSAD-CLIP: Zero-shot remote sensing anomaly detection of the Earth’s surface based on pre-trained vision-language model. InIGARSS 2025-2025 IEEE International Geoscience and Remote Sensing Sympo- sium, pages 1565–1569. IEEE, 2025. 3 on Pith
  14. [34] Zhuo Zheng, Ailong Ma, Liangpei Zhang, and Yanfei Zhong. Change is everywhere: Single-temporal supervised object change detection in remote sensing imagery. InProceedings of the IEEE/CVF International Conference on Computer Vi- sion (ICCV), pages 15193–15202, 2021. 2 on Pith
  15. [35] Zhuo Zheng, Shiqi Tian, Ailong Ma, Liangpei Zhang, and Yanfei Zhong. Scalable multi-temporal remote sensing change data generation via simulating stochastic change pro- cess. In2023 IEEE/CVF International Conference on Com- puter Vision (ICCV), pages 21818–21827. IEEE, 2023. 2 on Pith
  16. [36] Zhuo Zheng, Yanfei Zhong, Liangpei Zhang, and Stefano Er- mon. Segment any change.Advances in Neural Information Processing Systems, 37:81204–81224, 2024. 2 on Pith
  17. [37] Zhuo Zheng, Stefano Ermon, Dongjun Kim, Liangpei Zhang, and Yanfei Zhong. Changen2: Multi-temporal re- mote sensing generative change foundation model.IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(2):725–741, 2025. 2 on Pith
  18. [38] Qihang Zhou, Guansong Pang, Yu Tian, Shibo He, and Jim- ing Chen. AnomalyCLIP: Object-agnostic prompt learning for zero-shot anomaly detection. InInternational Conference on Learning Representations, 2024. 3 10 SALT: State-Aligned Low-Rank Transfer for Single-Temporal Cross-Da... on Pith

Challenges and discussion

Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.

No comments yet.

Sign in to comment