Pith. sign in

REVIEW 4 major objections 6 minor 38 references

Heuristical Comparison of Vision Transformers Against Convolutional Neural Networks for Semantic Segmentation on Remote Sensing Imagery

T0 review · 4 major / 6 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 42.9M-parameter UNet with a three-part weighted loss reaches 73.4 mIoU on iSAID, surpassing larger transformer models.

desk verdict Useful empirical measurements undermined by a missing baseline and protocol mismatch; the 'loss boost' claim does not hold as written. read the letter →

arxiv 2411.09101 v2 pith:ATZE2RIQ submitted 2024-11-14 cs.CV cs.AI

Ashim Dahal , Saydul Akbar Murad , Nick Rahimi This is my paper · ORCID
classification cs.CVcs.AI
keywords VisionTransformersemanticsegmentationremotesensingiSAIDUNetcombinedweightedlossmeanIntersectionoverUniontransferlearning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper reports an empirical comparison between a from-scratch UNet CNN and a transfer-learned MaskFormer vision transformer on the iSAID remote-sensing segmentation benchmark. It argues that the key driver of performance is not the encoder architecture alone but the loss function: a weighted sum of an IoU loss, a Dice loss, and a weighted cross-entropy term. With that combined loss, a 42.9M-parameter UNet reaches 73.4 mIoU on the validation set, a score the paper says surpasses published results from larger models such as the 113.8M-parameter AerialFormer-B (69.3 mIoU) and a 100M-parameter plain ViT (67.2 mIoU). The 200M-parameter MaskFormer baseline still scores higher in raw accuracy (82.48 mIoU), but at roughly five times the parameters and slower inference. A sympathetic reader would take the paper's central claim to be that a small CNN with a carefully weighted loss can approach or beat much larger attention-based models on this task, which matters for deployment on platforms with limited compute.

What carries the argument

The load-bearing object is the combined weighted loss function in Eq. (5): $$L = \lambda_{\mathrm{iou}} L_{\mathrm{iou}} + \lambda_{\mathrm{dice}} L_{\mathrm{dice}} + \lambda_{\mathrm{ce}} L_{\mathrm{ce}}$$ where $L_{\mathrm{iou}}$ replaces the non-differentiable intersection and union with multiplication and addition, $L_{\mathrm{dice}}$ uses the generalized Dice overlap, and $L_{\mathrm{ce}}$ is a weighted cross-entropy with background weight $\beta = 0.15$ and foreground weight 1. The weights $\lambda_{\mathrm{iou}} = 0.8$, $\lambda_{\mathrm{dice}} = 1$, and $\lambda_{\mathrm{ce}} = 10$ were chosen by trial and error. This loss is what the paper credits with letting the small UNet generalize on background pixels and small objects, while the UNet itself is a standard four-skip-connection encoder-decoder with 42.9M parameters. The MaskFormer baseline uses the same Dice-plus-focal loss from its reference implementation.

What would settle it

Evaluate the trained UNet on the official iSAID test set, or re-evaluate the cited baselines on the same validation split with the same background-excluded protocol; if the matched mIoU drops below the 69.3 reported for the 113.8M-parameter baseline, the central claim fails.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the proposed combined weighted loss function—$L = \lambda_{\mathrm{iou}} L_{\mathrm{iou}} + \lambda_{\mathrm{dice}} L_{\mathrm{dice}} + \lambda_{\mathrm{ce}} L_{\mathrm{ce}}$ with weights 0.8, 1, and 10—lets a generic 42.9M-parameter UNet, trained from scratch with mixed precision and gradient accumulation, reach 73.4 mIoU on the iSAID validation set while excluding the background class from evaluation. The same table reports per-class IoU scores that beat the cited AerialFormer-B, RingMo, and plain-ViT baselines on most categories, and the paper notes that the MaskFormer ViT (about 200M parameters) reaches 82.48 mIoU. The authors interpret this as evidence that the combined loss improves generalization and background handling, that the UNet is more inference-efficient (0.19s versus 0.29s on six images), and that the parameters needed for effective remote-sensing segmentation do not need to exceed 50M. They also state explicitly that background pixels are not counted in any of the evaluations, theirs or the prior works they compare against.

Load-bearing premise

The comparison assumes that the paper's validation-set mIoU, computed over 15 foreground classes only, is directly comparable to the test-set mIoU of the cited models, even though the cited works may use a different evaluation protocol or include the background class.

Editorial extensions

If this is right

  • If the 73.4 mIoU holds under matched evaluation, a 42.9M-parameter CNN can surpass published iSAID results from models two to five times larger, showing that loss design can compensate for a much smaller architecture.
  • The MaskFormer ViT's higher raw score (82.48 mIoU) at 200M parameters and slower inference shapes the trade-off as roughly five times the parameters for about 9 mIoU points.
  • The paper's inference measurements (0.19s versus 0.29s on six images) place the CNN ahead for time-critical applications such as UAV and ground-vehicle segmentation.
  • The paper's conclusion that a workable remote-sensing segmentation model does not need more than 50M parameters follows directly if the 73.4 mIoU result is reproducible under matched conditions.

Reading between the lines

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

  • The headline comparison is only as strong as its evaluation protocol: the 73.4 mIoU is a validation-set number with the background class excluded, whereas the cited works' numbers may come from the official test server with different background handling; a same-protocol rerun could shift the ranking.
  • The very low background weight ($\beta = 0.15$) in the cross-entropy term suggests the loss is engineered to ignore the unlabeled majority class; on fully labeled datasets with no background class, the weight balance would likely need to be re-derived.
  • A natural testable extension is to apply the same combined loss to other aerial or satellite segmentation benchmarks with different class imbalance, to see whether the gain is specific to iSAID or generalizes.
  • Because the three loss weights were selected by trial and error, the paper leaves open whether a small grid search or a learned weighting would yield further gains on the same CNN.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper presents a heuristic comparison between a custom 42.9M-parameter UNet CNN and a fine-tuned MaskFormer (Swin-L, ~200M parameters) on the iSAID remote-sensing semantic segmentation benchmark. The authors propose a weighted combined loss function (Equation 5) that sums IoU loss, Dice loss, and weighted cross-entropy loss, with weights λ_iou=0.8, λ_dice=1, and λ_ce=10. They report validation-set mIoU values of 73.4 for the UNet and 82.48 for the MaskFormer model, compare these against published test-set results, and claim that the combined loss significantly boosts the CNN's performance. The paper also compares inference time and parameter counts of the two models.

Significance. If the central claims were properly supported, the paper would offer a practically useful result: a relatively small CNN with a tuned loss approaching or exceeding larger ViT-based models on a difficult remote-sensing segmentation benchmark, with efficiency advantages. The manuscript provides a public code repository, describes the data augmentation pipeline in algorithmic detail, and reports parameter counts, FLOPs, and inference times, which are useful for reproducibility. However, the main empirical claims are not supported by the experiments as reported. The absence of a baseline UNet trained with a standard loss means the effect of the proposed loss is not identified, and the comparison to prior work is based on validation-set numbers versus test-set numbers without protocol reconciliation. The internal inconsistency between Table I and Figure 4 further undermines confidence in the reported metrics. These issues are load-bearing for the paper's headline conclusions, so the significance of the work is not established in its current form.

major comments (4)
  1. [Section IV and Section III-D (Equation 5)] The central claim that the proposed combined weighted loss significantly boosts the UNet's performance is not supported because no control experiment is reported. The paper never trains the same 42.9M UNet architecture with a standard cross-entropy, Dice, or IoU loss under the identical augmentation, optimizer, gradient-clipping, and evaluation protocol. Without such a baseline, the reported 73.4 mIoU cannot be attributed to Equation (5); architecture choices, augmentation, gradient accumulation, mixed precision, or hyperparameter tuning could equally explain the result.
  2. [Section IV, Table I, Section III-A] The comparison against prior state-of-the-art models is not valid as stated. The paper's mIoU values are computed on the validation set with the background class excluded (Section III-A: 'We only consider the foreground category while calculating our validation metrics scores'; Section IV: 'Our UNet CNN model's metrics on the validation set surpassed...'), while the cited results (e.g., AerialFormer, RingMo) are test-set results from their respective papers. Section IV asserts that 'the background class is not taken into consideration in any, ours or the previous, of the evaluation' without providing evidence for this claim. Unless the exact evaluation protocol (dataset split, background handling, class averaging, and whether previous works used the same validation set) is reconciled, the headline comparisons—UNet 73.4 versus AF-B 69.3, and MaskFormer 82.48 versus the cited models—are not established.
  3. [Section IV versus Table I] There is an internal inconsistency in the reported UNet accuracy. Table I reports an mIoU of 73.4 for 'Ours (CNN)', while Section IV states that the UNet 'reached the peak of 81% and stabilized at around 78% during training over all classes' and refers to Figure 4. The paper must clarify whether Table I reports the final-epoch validation mIoU over the 15 foreground classes only, and why Figure 4 shows different peak and final values. As written, the two numbers cannot both describe the same metric, and this discrepancy undermines the reliability of the reported results.
  4. [Section III-D and Section III-F] The loss-function hyperparameters (λ_iou=0.8, λ_dice=1, λ_ce=10) and the class weight β=0.15 are selected by trial and error on the same validation set used for reporting the headline results, and no ablation or sensitivity analysis is provided. Since Equation (5) is the paper's main contribution, the lack of any ablation leaves both the contribution and the risk of validation-set overfitting unquantified. A minimal analysis varying each weight while holding the others fixed is needed to demonstrate that the loss components matter and that the specific values are not the sole driver of the reported mIoU.
minor comments (6)
  1. [Section II] Several citations are unresolved placeholders, including '[?]' after 'Papers with Code' and after 'SegFormer' and 'MaskFormer' in the related-work discussion; these need to be filled in before submission.
  2. [Algorithm 2] The gradient clipping update rule is typeset incorrectly; it should read: if ||ĝ|| > threshold, set ĝ ← (threshold / ||ĝ||) · ĝ.
  3. [Section III-D, Equation (4)] The description of β is ambiguous: the text says 'β is the weight hyperparameter (0.15 for unlabelled class and 1 for the rest)' but then says 'We choose the minimal value of 0.15 for β.' Please clarify whether β is a scalar or a per-class weight vector, and specify the per-pixel formula when β=1 for foreground classes; as written, β=1 in Equation (4) makes the second term vanish for foreground pixels, which should be stated explicitly.
  4. [Introduction] The sentence 'One of the main pitfalls of such datasets is the background class [11]' cites a qualitative health research paper (Hinds et al.) that is not relevant to background-class handling in segmentation; an appropriate citation on class-imbalance or background dominance in remote-sensing datasets should be provided.
  5. [Section IV] The phrase 'This correlation fgvrther validates the experimental design' contains a typo ('fgvrther' should be 'further'), and the causal wording 'validates the experimental design of the combined loss' overstates what visual inspection of qualitative segmentation maps can establish.
  6. [Table I and Section III-C] The MaskFormer row lists 200M parameters, while the text says 'resulting to approximately 200M parameters'; please report the exact parameter count and state clearly how many parameters are trainable versus frozen, since the fine-tuning protocol is part of the comparison.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the combined loss is an explicit weighted sum of standard losses, and the reported mIoU is an empirical measurement rather than a quantity derived by construction from the fitted weights.

full rationale

The paper's central claim is that a weighted fused loss (Eq. 5) improves UNet segmentation. The loss is an explicit weighted sum of standard IoU, Dice, and cross-entropy terms (Eqs. 2-4), each externally cited; no term is defined in terms of the reported mIoU. The 73.4 mIoU is an empirical validation-set measurement after training, not an algebraic consequence of the fitted lambda values, so no equation reduces the result to the loss weights by construction. The trial-and-error selection of lambda_iou=0.8, lambda_dice=1, lambda_ce=10 and beta=0.15, together with measuring the headline mIoU on the same validation set, is a selection-bias / evaluation-leakage concern, not circularity: the mIoU is not statistically forced to equal a function of the lambdas. Comparing validation-set mIoU to prior test-set mIoU and asserting without evidence that prior work also excludes background are comparability flaws, not self-referential derivations. No load-bearing self-citations exist: MaskFormer, UNet, IoU loss, Dice loss, and cross-entropy are all external references. The paper explicitly discloses the trial-and-error tuning, the background-exclusion protocol, and the test-image de-segmentation caveat, which further supports treating these as methodological limitations. Under the strict reduction requirement, no circular step can be exhibited.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim depends on several hand-fitted loss weights and evaluation-protocol assumptions. No fundamentally new entity is introduced; the 'novel' loss is a weighted sum of existing losses, and the reported improvement is tied to parameter values tuned on the same validation set used for the final scores.

free parameters (6)
  • λ_iou = 0.8
    Weight for soft-IoU loss term, tuned by trial and error on the validation set (Section III-D).
  • λ_dice = 1
    Weight for Dice loss term, tuned by trial and error on the validation set (Section III-D).
  • λ_ce = 10
    Weight for cross-entropy loss term, tuned by trial and error on the validation set (Section III-D).
  • β (background class weight in Lce) = 0.15
    Hand-chosen class weight for unlabelled pixels, with weight 1 for foreground classes (Section III-D).
  • Gradient clipping threshold = 3.0
    Maximum gradient norm, chosen by trial and error (Section III-C).
  • Training settings (learning rate, epochs, augmentation fractions) = 1e-3; 40 epochs; crop 6-28%; flip 50%; rotation 0-360
    Chosen by hand without systematic search (Sections III-B, III-F).
assumptions (5)
  • domain assumption The iSAID validation set is representative of the test distribution
    The paper uses validation mIoU as the headline result and compares it to prior test-set results without noting any mismatch (Section IV).
  • domain assumption Prior published mIoU values were computed over the same 15 foreground classes
    The paper states the background class is not taken into consideration in any evaluation, ours or the previous (Section IV), but provides no evidence for prior protocols.
  • standard math Soft IoU and Dice formulations in Eq. (2) and (3) are valid differentiable surrogates for the true metrics
    Standard relaxation of nondifferentiable intersection and union operations (Section III-D).
  • domain assumption Standard supervised training on the train split with the chosen loss generalizes to the validation distribution
    Empirical deep learning assumption underlying the entire evaluation (Sections III, IV).
  • domain assumption Adam with gradient clipping and mixed precision does not materially distort the metric comparison
    Both models are trained with the same optimizer and clipping, but no sensitivity analysis is provided (Sections III-C, III-F).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Heuristical Comparison of Vision Transformers Against Convolutional Neural Networks for Semantic Segmentation on Remote Sensing Imagery." pith.science (2026). https://pith.science/paper/ATZE2RIQ

@misc{pith2026241109101,
  author       = {Pith},
  title        = {Pith review of: Heuristical Comparison of Vision Transformers Against Convolutional Neural Networks for Semantic Segmentation on Remote Sensing Imagery},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ATZE2RIQ}},
  note         = {Machine review of arXiv:2411.09101}
}
read the original abstract

Vision Transformers (ViT) have recently brought a new wave of research in the field of computer vision. These models have performed particularly well in image classification and segmentation. Research on semantic and instance segmentation has accelerated with the introduction of the new architecture, with over 80% of the top 20 benchmarks for the iSAID dataset based on either the ViT architecture or the attention mechanism behind its success. This paper focuses on the heuristic comparison of three key factors of using (or not using) ViT for semantic segmentation of remote sensing aerial images on the iSAID dataset. The experimental results observed during this research were analyzed based on three objectives. First, we studied the use of a weighted fused loss function to maximize the mean Intersection over Union (mIoU) score and Dice score while minimizing entropy or class representation loss. Second, we compared transfer learning on Meta's MaskFormer, a ViT-based semantic segmentation model, against a generic UNet Convolutional Neural Network (CNN) based on mIoU, Dice scores, training efficiency, and inference time. Third, we examined the trade-offs between the two models in comparison to current state-of-the-art segmentation models. We show that the novel combined weighted loss function significantly boosts the CNN model's performance compared to transfer learning with ViT. The code for this implementation can be found at: https://github.com/ashimdahal/ViT-vs-CNN-Image-Segmentation.

Figures

Figures reproduced from arXiv: 2411.09101 by the authors.

Figure 1
Figure 1. Brief Overview of the Training and Validation Lifecycle [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Custom UNet CNN architecture [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. MaskFormer ViT architecture from [2] We also use a weighted cross-entropy loss function to maintain the entropy of our predictions, so the third part of our function becomes Lce [16]. Lce = β ∗ A log(B) + (1 − β)(1 − A) log(1 − B) (4) where β is the weight hyperparameter (0.15 for unlabelled class and 1 for the rest), Combining the loss functions (2) (3) (4) together with weights λiou, λdice, λce respectively we get… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Comparison of Metrics over Epochs We choose the minimal value of 0.15 for β in Lce. This loss function in Eq. 5 represents the final form of our proposed loss. E. Validation Metrics We validate the results produced by our model for C classes using the mIoU score and Di…
Figure 5
Figure 5. Figure 5: Sample Visualization of model’s output First, we list out the two model’s efficiency information in Table II. During training, most of the time was consumed by the data augmentation technique described in III-B, which took 65 seconds per batch for a batch size of 128. …
Figure 6
Figure 6. Figure 6: Class-wise comparison of IoU and Dice Scores [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 30 canonical work pages

  1. [1]

    SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation,

    V . Badrinarayanan, A. Kendall, and R. Cipolla, “SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 39, no. 12, pp. 2481–2495, 2017

    work page 2017

  2. [2]

    Per-Pixel Classification is Not All You Need for Semantic Segmentation,

    B. Cheng, A. Schwing, and A. Kirillov, “Per-Pixel Classification is Not All You Need for Semantic Segmentation,” in Advances in Neural Information Processing Systems , vol. 34, pp. 17864–17875, 2021

    work page 2021

  3. [3]

    ImageNet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “ImageNet: A large-scale hierarchical image database,” in 2009 IEEE Conference on Computer Vision and Pattern Recognition , pp. 248–255, 2009

    2009

  4. [4]

    ResUNet- a: A deep learning framework for semantic segmentation of remotely sensed data,

    F. I. Diakogiannis, F. Waldner, P. Caccetta, and C. Wu, “ResUNet- a: A deep learning framework for semantic segmentation of remotely sensed data,” ISPRS Journal of Photogrammetry and Remote Sensing , vol. 162, pp. 94–114, 2020. 10

    work page 2020

  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,” in International Conference on Learning Representations, 2021

    work page 2021

  6. [6]

    Is Attention Better Than Matrix Decomposition?,

    Z. Geng, M.-H. Guo, H. Chen, X. Li, K. Wei, and Z. Lin, “Is Attention Better Than Matrix Decomposition?,” in International Conference on Learning Representations, 2021

    work page 2021

  7. [7]

    SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation,

    M.-H. Guo, C.-Z. Lu, Q. Hou, Z. Liu, M.-M. Cheng, and S.-M. Hu, “SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation,” arXiv preprint arXiv:2209.08575 , 2022

    arXiv 2022

  8. [8]

    Visual Attention Network,

    M.-H. Guo, C.-Z. Lu, Z.-N. Liu, M.-M. Cheng, and S.-M. Hu, “Visual Attention Network,” arXiv preprint arXiv:2202.09741 , 2022

    arXiv 2022

Show all 38 references
  1. [9]

    AerialFormer: Multi-Resolution Transformer for Aerial Image Segmentation,

    T. Hanyu et al. , “AerialFormer: Multi-Resolution Transformer for Aerial Image Segmentation,” Remote Sensing , vol. 16, no. 16, 2024

    2024

  2. [10]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition , pp. 770–778, 2016

    2016

  3. [11]

    The possibilities and pitfalls of doing a secondary analysis of a qualitative data set,

    P. S. Hinds, R. J. V ogel, and L. Clarke-Steffen, “The possibilities and pitfalls of doing a secondary analysis of a qualitative data set,” Qualitative health research , vol. 7, no. 3, pp. 408–424, 1997

    1997

  4. [12]

    Weakly-supervised learning based automatic augmentation of aerial insulator images,

    D. Jiang, Y . Cao, and Q. Yang, “Weakly-supervised learning based automatic augmentation of aerial insulator images,” Expert Systems with Applications , vol. 242, p. 122739, 2024

    2024

  5. [13]

    Adam: A method for stochastic optimization,

    D. P. Kingma, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  6. [14]

    Focal Loss for Dense Object Detection,

    T. Lin, “Focal Loss for Dense Object Detection,” arXiv preprint arXiv:1708.02002, 2017

    2017 arXiv

  7. [15]

    Swin transformer: Hierarchical vision transformer using shifted windows,

    Z. Liu et al. , “Swin transformer: Hierarchical vision transformer using shifted windows,” in Proceedings of the IEEE/CVF international conference on computer vision , pp. 10012–10022, 2021

    2021

  8. [16]

    Cross-entropy loss functions: Theoretical analysis and applications,

    A. Mao, M. Mohri, and Y . Zhong, “Cross-entropy loss functions: Theoretical analysis and applications,” in International conference on Machine learning , pp. 23803–23828, 2023

    2023

  9. [17]

    On the difficulty of training recurrent neural networks,

    R. Pascanu, T. Mikolov, and Y . Bengio, “On the difficulty of training recurrent neural networks,” in Proceedings of the 30th International Conference on Machine Learning , vol. 28, no. 3, pp. 1310–1318, 2013

    2013

  10. [18]

    Regmi, Unsupervised Image Segmentation in Satellite Imagery Using Deep Learning

    S. Regmi, Unsupervised Image Segmentation in Satellite Imagery Using Deep Learning . The University of Alabama in Huntsville, 2023

    2023

  11. [19]

    U-Net: Convolutional Networks for Biomedical Image Segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” in Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 , pp. 234–241, 2015

    2015

  12. [20]

    Very Deep Convolutional Networks for Large-Scale Image Recognition,

    K. Simonyan and A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” CoRR, vol. abs/1409.1556, 2014

    2014 arXiv

  13. [21]

    Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations,

    C. H. Sudre, W. Li, T. Vercauteren, S. Ourselin, and M. Jorge Car- doso, “Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations,” in Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support , pp. 240–...

    2017

  14. [22]

    RingMo: A Remote Sensing Foundation Model With Masked Image Modeling,

    X. Sun et al. , “RingMo: A Remote Sensing Foundation Model With Masked Image Modeling,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–22, 2023

    2023

  15. [23]

    Attention is all you need,

    A. Vaswani et al., “Attention is all you need,” inProceedings of the 31st International Conference on Neural Information Processing Systems , pp. 6000–6010, 2017

    2017

  16. [24]

    Advancing plain vision transformer toward remote sensing foundation model,

    D. Wang et al. , “Advancing plain vision transformer toward remote sensing foundation model,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–15, 2022

    2022

  17. [25]

    Samrs: Scaling-up remote sensing segmentation dataset with segment anything model,

    D. Wang, J. Zhang, B. Du, M. Xu, L. Liu, D. Tao, and L. Zhang, “Samrs: Scaling-up remote sensing segmentation dataset with segment anything model,” Advances in Neural Information Processing Systems , vol. 36, 2024

    2024

  18. [26]

    Transformers: State-of-the-Art Natural Language Pro- cessing,

    T. Wolf et al. , “Transformers: State-of-the-Art Natural Language Pro- cessing,” in Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations , pp. 38–45, 2020

    2020

  19. [27]

    W-Net: Convolutional neural network for segmenting remote sensing images by dual path semantics,

    G. Liu, Q. Wang, J. Zhu, and H. Hong, “W-Net: Convolutional neural network for segmenting remote sensing images by dual path semantics,” PLOS ONE , vol. 18, no. 7, pp. 1–16, 2023

    2023

  20. [28]

    AANet: an attention- based alignment semantic segmentation network for high spatial resolu- tion remote sensing images,

    G. Xue, Y . Liu, Y . Huang, M. Li, and G. Yang, “AANet: an attention- based alignment semantic segmentation network for high spatial resolu- tion remote sensing images,” International Journal of Remote Sensing , vol. 43, no. 13, pp. 4836–4852, 2022

    2022

  21. [29]

    DOTA: A Large-Scale Dataset for Object Detection in Aerial Images,

    G.-S. Xia et al. , “DOTA: A Large-Scale Dataset for Object Detection in Aerial Images,” in The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2018

    2018

  22. [30]

    Unified perceptual parsing for scene understanding,

    T. Xiao, Y . Liu, B. Zhou, Y . Jiang, and J. Sun, “Unified perceptual parsing for scene understanding,” in Proceedings of the European conference on computer vision (ECCV) , pp. 418–434, 2018

    2018

  23. [31]

    TreeUNet: Adaptive Tree convolutional neural networks for subdecimeter aerial image segmentation,

    K. Yue, L. Yang, R. Li, W. Hu, F. Zhang, and W. Li, “TreeUNet: Adaptive Tree convolutional neural networks for subdecimeter aerial image segmentation,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 156, pp. 1–13, 2019

    2019

  24. [32]

    iSAID: A Large-scale Dataset for Instance Seg- mentation in Aerial Images,

    S. Waqas Zamir et al., “iSAID: A Large-scale Dataset for Instance Seg- mentation in Aerial Images,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops , pp. 28–37, 2019

    2019

  25. [33]

    Context Encoding for Semantic Segmentation,

    H. Zhang, K. J. Dana, J. Shi, Z. Zhang, X. Wang, A. Tyagi, and A. Agrawal, “Context Encoding for Semantic Segmentation,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 7151–7160, 2018

    2018

  26. [34]

    Vitaev2: Vision transformer advanced by exploring inductive bias for image recognition and be- yond,

    Q. Zhang, Y . Xu, J. Zhang, and D. Tao, “Vitaev2: Vision transformer advanced by exploring inductive bias for image recognition and be- yond,” International Journal of Computer Vision , vol. 131, no. 5, pp. 1141–1162, 2023

    2023

  27. [35]

    Iou loss for 2d/3d object detection,

    D. Zhou, J. Fang, X. Song, C. Guan, J. Yin, Y . Dai, and R. Yang, “Iou loss for 2d/3d object detection,” in 2019 international conference on 3D vision (3DV) , pp. 85–94, 2019

    2019

  28. [36]

    MultiResUNet : Rethinking the U-Net architecture for multimodal biomedical image segmentation,

    N. Ibtehaz and M. S. Rahman, “MultiResUNet : Rethinking the U-Net architecture for multimodal biomedical image segmentation,” Neural Networks, vol. 121, pp. 74–87, 2020. [Previous entries from Pascanu to Noh...]

    2020

  29. [37]

    Learning Deconvolution Network for Semantic Segmentation,

    H. Noh, S. Hong, and B. Han, “Learning Deconvolution Network for Semantic Segmentation,” in 2015 IEEE International Conference on Computer Vision (ICCV), pp. 1520–1528, 2015. [Previous entries from Regmi to Zamir...]

    2015

  30. [38]

    Using GAN Methods for Aerial Images Segmentation,

    S.ALTUNG ¨UVEN and B. TOPTAS ¸, “Using GAN Methods for Aerial Images Segmentation,” Dicle University Journal of Engineering , vol. 15, no. 1, 2024

    2024

Pith tools

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