Pith. sign in

REVIEW 4 major objections 6 minor 44 references

Bridging Classical and Modern Computer Vision: PerceptiveNet for Tree Crown Semantic Segmentation

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read PerceptiveNet, a segmentation network whose first layer uses trainable Log-Gabor filters, reports the highest mean intersection-over-union scores on three aerial datasets, including a 48.1% mIoU on TreeCrown.

desk verdict A plausible Log-Gabor backbone with a clean ablation, but the SOTA claim rests on single-run mIoU gaps and an underspecified split protocol. read the letter →

arxiv 2505.23597 v1 pith:VAKWMOTF submitted 2025-05-29 cs.CV

Georgios Voulgaris This is my paper
classification cs.CV
keywords treecrownsegmentationLog-GaborfilterssemanticaerialimagerydeeplearningUAVremotesensingCNNtransformerhybrid
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 argues that a convolutional layer parameterised by trainable Log-Gabor functions, inserted as the first layer of a ResUNet-style encoder-decoder, gives the network a more balanced frequency coverage and better shape-based features, and that combining this layer with mixed max-average pooling and averaged dilated convolutions yields a backbone that segments tree crowns in dense aerial forest imagery more accurately than standard U-Net variants and transformer hybrids. On the TreeCrown dataset the complete model, PerceptiveNet, reaches 48.1% mean IoU, which the paper reports as a 10.5 percentage-point improvement over its ResUNet baseline and above the other compared CNN models; the same backbone also posts the highest mIoU on Landcover.AI (81.6%) and UA Vid (68.6%). The paper further claims that the backbone transfers to a hybrid CNN-Transformer model, PerceptiveNeTr, which outperforms pure-transformer comparators. A sympathetic reader would care because tree-crown segmentation is the basis for forest health, biodiversity, and carbon accounting, and the proposed layer is a small architectural change that could be adopted wherever aerial imagery suffers from shadows, scale variation, and weak spectral differences.

What carries the argument

The load-bearing object is the Log-Gabor-parameterised convolutional layer: a 2D filter $g(x,y) = g_r(r) \cdot g_\theta(\theta) \cdot \cos(2\pi f_0 r + \psi) / (2\pi \sigma^2)$, with radial and angular Gaussian envelopes in the log-frequency domain, whose parameters ($f, \theta, \sigma, \psi, f_0, \theta_0$) are initialised and then learned by backpropagation. Its stated theoretical advantages are a zero DC component, better orthogonality across scales, uniform Fourier-domain coverage, and better spatial localisation compared with ordinary Gabor filters. The layer is combined with two supporting mechanisms: a mixture of maximum and average pooling with mixing ratio $\alpha = 0.8$ to widen the receptive field while halving resolution, and an averaged dilated convolutional block that sums the localisation maps from dilation rates 1, 3, 6, 9 to avoid spurious high-rate responses. Together these form the PerceptiveNet residual encoder; the decoder mirrors the residual block with the dilated layer. The hybrid PerceptiveNeTr appends four transformer layers (patch embedding, LayerNorm, multi-head self-attention, MLP) after the PerceptiveNet encoder.

What would settle it

Run each compared model at least five times with different random seeds and the same data split, then check whether PerceptiveNet's mIoU advantage over the closest baseline (e.g., MACUNet on TreeCrown) exceeds the standard deviation of the runs; additionally, re-split the data by orthophoto so that no patches from the same orthophoto appear in both train and test, and see whether the reported margins survive.

Watch

Extended reading notes

Core claim

The central claim is that replacing the initial convolutional layer of a ResUNet with a layer whose filter weights are generated by a Log-Gabor function with trainable parameters—frequency, orientation, bandwidth, phase—and pairing that with residual blocks that use a mixture of maximum and average pooling and an averaged set of dilated convolutions (rates 1, 3, 6, 9) produces a segmentation model that captures both fine spatial detail and broad context. The paper reports PerceptiveNet achieving the highest mIoU of all compared models on all three datasets: 48.1% on TreeCrown, 81.6% on Landcover.AI, and 68.6% on UA Vid, with ablation results attributing gains to the Log-Gabor layer, the pooling mixture, and the dilated block in combination. It also reports that the same backbone inside a CNN-Transformer hybrid, PerceptiveNeTr, reaches 42.0%, 75.3%, and 64.3% mIoU on the three datasets, exceeding pure-transformer UNETR 2D and SwinUNet, and that Class Activation Maps show more focused activation on tree regions than the ResUNet baseline.

Load-bearing premise

The evaluation treats single-run mIoU differences as meaningful; the load-bearing premise is that a gap of a few points between models reflects real architectural superiority rather than run-to-run noise, and that the test patches do not share scenes with training patches.

Editorial extensions

If this is right

  • If the reported scores hold, PerceptiveNet would set the highest mIoU on TreeCrown, Landcover.AI, and UA Vid among the models compared, making the Log-Gabor first layer a drop-in alternative for aerial segmentation pipelines.
  • The ablation suggests the three components are complementary: removing any one lowers mIoU, so practitioners should keep all three rather than choosing one.
  • The backbone's success inside PerceptiveNeTr implies the benefit of Log-Gabor features is not tied to a pure-CNN decoder, but persists when global context is added by transformers.
  • The qualitative results on dead and overlapping trees suggest the model is better at high-frequency boundary detail, which is exactly where dense-forest canopies fail.

Reading between the lines

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

  • Editorial inference: if the Log-Gabor first layer is the true source of the gain, it should transfer to other encoder-decoder backbones (e.g., DeepLabV3+) when swapped into the first convolution; the paper does not test this, but it is a cheap experiment.
  • Editorial inference: the zero-DC property of Log-Gabor filters suggests the layer acts as a learned band-pass front-end; one could test this directly by visualising the learned filters' frequency responses after training to see whether they spread across scales as the theory assumes.
  • Editorial inference: the dataset's split method (by orthophoto vs by random crop) is not reported; if patches from the same orthophoto appear in both train and test, the reported mIoU could be inflated by spatial autocorrelation, so the strongest follow-up would evaluate on held-out orthophotos.
  • Editorial inference: because the paper reports only single-run results, the margins over the nearest baselines may not be stable; a prudent next step is to rerun each model with several seeds and report mean and standard deviation.
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 proposes PerceptiveNet, a U-Net-style segmentation architecture whose initial convolution is parameterized by trainable Log-Gabor filters, and which incorporates residual blocks with mixed max-average pooling and averaged dilated convolutions. It also introduces PerceptiveNeTr, a hybrid CNN-Transformer variant using PerceptiveNet as the backbone. The authors compare against ResUNet, U-Net, MACUNet, DeepLabV3+, UNETR 2D, and SwinUNet on three aerial datasets (TreeCrown, Landcover.AI, UA Vid), reporting mIoU improvements of roughly 2.5 to 10.5 points over the strongest baselines, with an ablation study that decomposes the contributions of the proposed components. The central claim is that PerceptiveNet outperforms all compared models on all three datasets and that the Log-Gabor layer provides a consistent gain over standard and Gabor-based convolutions.

Significance. If the reported results are reproducible, the work offers a simple and well-motivated architectural modification—trainable Log-Gabor filters—that improves segmentation on aerial forest and urban scenes, with a clear theoretical rationale based on zero DC components and more uniform frequency coverage. The multi-dataset evaluation and the ablation isolating pooling, dilated convolution, and Log-Gabor effects are useful contributions. However, the empirical evidence is currently thin: all numbers come from single training runs, no confidence intervals or significance tests are provided, the train/test split protocol for the two benchmark datasets is not described, and key hyperparameters are hand-picked. Because the headline margins are only a few mIoU points, the claimed state-of-the-art status is not yet established. The paper does not provide code or trained models, which further limits reproducibility.

major comments (4)
  1. [§3.4, Tables 3 and 4] The central claim that PerceptiveNet outperforms all compared models rests entirely on single-run mIoU values. For example, the tree-crown margin over MACUNet is 2.2 mIoU points (48.1 vs. 45.9), the Landcover.AI margin over DeepLabV3+ is 3.0 points (81.6 vs. 78.6), and the UA Vid margin over MACUNet is 2.5 points (68.6 vs. 66.1). Training from scratch with Adam and no stated learning-rate schedule can produce run-to-run variation of several mIoU points. Please report mean and standard deviation over at least three to five seeds, and provide a significance test or an explicit statement of variance; otherwise the reported ranking may not be reproducible.
  2. [§2.2, Data] The paper does not state whether the Landcover.AI and UA Vid splits are performed by source orthophoto/video or by random crop. Landcover.AI is sub-sampled from 33+8 large orthophotos into 21,924 crops; if crops from the same orthophoto appear in both training and test sets, spatially correlated pixels can inflate mIoU. Similarly, UA Vid is cropped into eight patches per original frame, and a random patch-level split can place patches from the same frame in both training and test. TreeCrown has repeated monthly acquisitions of the same stands, so a random image-level split may leak the same crowns across the boundary. Please specify the split protocol explicitly and, ideally, re-run with source-image/frame-level splits to rule out leakage.
  3. [§2.1, Model Training and hyperparameters] The mixing ratio α=0.8, the dilation rates 1, 3, 6, 9, the four transformer layers, and the 130 training epochs are presented as fixed choices without sensitivity analysis. Since these values were apparently selected using the validation sets before reporting test scores, and since only one configuration is evaluated, the reported differences may reflect selection bias rather than a robust architectural advantage. Please provide an ablation over α and the dilation-rate setting, or at least report results for multiple seeds under the chosen configuration, to show that the conclusion is not an artifact of a single hand-picked configuration.
  4. [§2.1, PerceptiveNeTr and Figure 4] The hybrid CNN-Transformer model is described only at a high level: patch embedding, LayerNorm, MSA, and MLP are mentioned, but the patch size, embedding dimension, number of heads, MLP hidden dimension, and positional encoding are not specified. Without these details, the transformer comparison in Table 4 cannot be reproduced, and the reported gains of PerceptiveNeTr over ViTResUNet cannot be attributed to the backbone rather than to differences in the transformer configuration. Please add a complete specification of the transformer hyperparameters.
minor comments (6)
  1. [§2.1, Eqs. (4)–(6)] There is a notation inconsistency in the Log-Gabor layer: the text lists f as a learnable parameter, but Eq. (4) and Eq. (5) use only f0 and σ in the radial and cosine terms, and no equation contains f. Please clarify whether f is a separate learnable parameter or a typo, and define all symbols in one place.
  2. [§2.2 and throughout] The dataset name 'UA Vid' is written inconsistently; the original source is 'UAVid'. Please use the official name for consistency and to aid readers in locating the dataset.
  3. [Table 4 and §3.4.2] Calling ViTResUNet a 'Proposed Hybrid CNN-Transformer' is misleading because ViTResUNet is the ResUNet-backbone baseline used to isolate the effect of the PerceptiveNet backbone. Please rephrase to avoid implying that ViTResUNet is a contribution.
  4. [§3.5 and Figure 6] The qualitative claims about boundary quality and dead-tree segmentation are plausible but not quantified. Consider reporting a boundary-oriented metric (e.g., boundary IoU or contour F1) to support these visual observations.
  5. [References] Some references are incomplete; for example, [25] and [26] list only the title and venue without author names. Please complete the bibliography.
  6. [§2.1, Model Training] The learning rate is not reported, and it is unclear whether any learning-rate schedule is used. Since the paper notes that 130 epochs were chosen based on validation improvement, please state the exact training protocol, optimizer settings, and the criterion for stopping.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the paper is an empirical architecture comparison; the only author-overlapping citation is non-load-bearing, and no result reduces to its inputs by construction.

full rationale

The paper is an empirical architecture comparison, not a derivation, and no claimed result is equivalent to its inputs by construction. PerceptiveNet's mIoU values in Section 3.4 are reported after training and evaluation, and the Log-Gabor layer is a standard differentiable parameterisation from established filtering theory (Equations 3-6), with its theoretical advantages cited to external prior work such as Fischer et al. [28] and Nava et al. [23]. The ablation study isolates architectural components, so claims that the components complement one another are empirical outcomes rather than tautologies. Hyperparameters such as alpha = 0.8 and dilation rates d = 1, 3, 6, 9 were selected on validation data, which is routine model selection and does not amount to a fitted input renamed as a prediction. The only author-overlapping citation is reference [19] (Voulgaris et al.), used to support the premise that CNNs over-rely on texture in aerial data; this is a motivating observation and is not load-bearing for the headline state-of-the-art comparisons, which are evaluated against external baselines and datasets. Concerns about missing confidence intervals and unspecified train/test patch splitting are reproducibility or statistical-validity issues, not circularity. Score 2 reflects one minor, non-load-bearing self-citation; no circular derivation chain was found.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The primary contribution is architectural: a new layer and backbone. The hand-picked hyperparameters are modeled as free parameters because they influence the central result. No new physical or conceptual entities are introduced.

free parameters (5)
  • pooling mixing ratio α_l = 0.8
    Chosen by hand for all layers ('For the purpose of this work, we chose a scalar mixing portion αl = 0.8') with no ablation.
  • dilation rates = 1, 3, 6, 9
    Selected without justification; the impact of different rate sets is not studied.
  • number of transformer layers = 4
    Set in PerceptiveNeTr architecture description without ablation.
  • training epochs = 130
    Chosen based on when validation stopped improving; this is a model selection choice.
  • δ small constant = unspecified
    Introduced to prevent division by zero in Eq. (5), but its value is not reported.
assumptions (3)
  • standard math The Log-Gabor parameterization in Eqs. (4)-(6) is differentiable and compatible with gradient backpropagation through the convolutional layer.
    The paper asserts the parameters (f, θ, σ, ψ, f0, θ0) are learnable and updated during training, which requires differentiability. This is a standard assumption but not proven in the text.
  • domain assumption The preprocessed Landcover.AI and UA Vid patches are split into train/val/test without spatial leakage between patches from the same orthophoto.
    The paper describes sub-sampling orthophotos into patches but does not state whether the split is by orthophoto or random, so the results assume no leakage.
  • ad hoc to paper The hand-chosen hyperparameters (α=0.8, dilation rates 1/3/6/9, 4 transformer layers) generalize across the three datasets without further tuning.
    These values are fixed for all experiments, but there is no evidence they are optimal or transferable, and they may have been selected based on the same test datasets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bridging Classical and Modern Computer Vision: PerceptiveNet for Tree Crown Semantic Segmentation." pith.science (2026). https://pith.science/paper/VAKWMOTF

@misc{pith2026250523597,
  author       = {Pith},
  title        = {Pith review of: Bridging Classical and Modern Computer Vision: PerceptiveNet for Tree Crown Semantic Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VAKWMOTF}},
  note         = {Machine review of arXiv:2505.23597}
}
read the original abstract

The accurate semantic segmentation of tree crowns within remotely sensed data is crucial for scientific endeavours such as forest management, biodiversity studies, and carbon sequestration quantification. However, precise segmentation remains challenging due to complexities in the forest canopy, including shadows, intricate backgrounds, scale variations, and subtle spectral differences among tree species. Compared to the traditional methods, Deep Learning models improve accuracy by extracting informative and discriminative features, but often fall short in capturing the aforementioned complexities. To address these challenges, we propose PerceptiveNet, a novel model incorporating a Logarithmic Gabor-parameterised convolutional layer with trainable filter parameters, alongside a backbone that extracts salient features while capturing extensive context and spatial information through a wider receptive field. We investigate the impact of Log-Gabor, Gabor, and standard convolutional layers on semantic segmentation performance through extensive experimentation. Additionally, we conduct an ablation study to assess the contributions of individual layers and their combinations to overall model performance, and we evaluate PerceptiveNet as a backbone within a novel hybrid CNN-Transformer model. Our results outperform state-of-the-art models, demonstrating significant performance improvements on a tree crown dataset while generalising across domains, including two benchmark aerial scene semantic segmentation datasets with varying complexities.

Figures

Figures reproduced from arXiv: 2505.23597 by the authors.

Figure 1
Figure 1. Tree Crown Semantic Segmentation, depicting (from [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Dense forest canopy, demonstrating the impact of shad [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Building blocks of the proposed Architecture: (a) PerceptiveNet architecture, (b) Decoder proposed dilated residual unit (DilRes), [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: PerceptiveNeTr: Hybrid CNN (PerceptiveNet) - [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Class Activation Maps, ResUNet vs PerceptiveNet En [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Aerial Tree Crown Semantic Segmentation of a Dense Forest (TreeCrown), Comprised of Visually Similar Tree Species. The left [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

44 extracted references · 42 canonical work pages

  1. [1]

    The validation of the mixedwood growth model (mgm) for use in forest management decision making,

    M. Bokalo, K. J. Stadt, P. G. Comeau, and S. J. Titus, “The validation of the mixedwood growth model (mgm) for use in forest management decision making,” Forests, vol. 4, no. 1, pp. 1–27, 2013. 1

    work page 2013

  2. [2]

    Develop- ment of crown ratio and height to crown base models for masson pine in southern china,

    Y . Li, W. Wang, W. Zeng, J. Wang, and J. Meng, “Develop- ment of crown ratio and height to crown base models for masson pine in southern china,” Forests, vol. 11, no. 11, p. 1216, 2020. 1

    work page 2020

  3. [3]

    Tree crown delineation al- gorithm based on a convolutional neural network,

    J. R. G. Braga, V . Peripato, R. Dalagnol, M. P. Ferreira, Y . Tarabalka, L. E. OC Arag˜ao, H. F. de Campos Velho, E. H. Shiguemori, and F. H. Wagner, “Tree crown delineation al- gorithm based on a convolutional neural network,” Remote Sensing, vol. 12, no. 8, p. 1288, 2020. 1

    work page 2020

  4. [4]

    Uav re- mote sensing monitoring of pine forest diseases based on im- proved mask r-cnn,

    G. Hu, T. Wang, M. Wan, W. Bao, and W. Zeng, “Uav re- mote sensing monitoring of pine forest diseases based on im- proved mask r-cnn,” International Journal of Remote Sens- ing, vol. 43, no. 4, pp. 1274–1305, 2022. 1

    work page 2022

  5. [5]

    Tree extraction from multi-scale uav images us- ing mask r-cnn with fpn,

    N. E. Ocer, G. Kaplan, F. Erdem, D. Kucuk Matci, and U. Avdan, “Tree extraction from multi-scale uav images us- ing mask r-cnn with fpn,” Remote sensing letters , vol. 11, no. 9, pp. 847–856, 2020. 1

    work page 2020

  6. [6]

    An improved res-unet model for tree species classification using airborne high-resolution images,

    K. Cao and X. Zhang, “An improved res-unet model for tree species classification using airborne high-resolution images,” Remote Sensing, vol. 12, no. 7, p. 1128, 2020. 1

    work page 2020

  7. [7]

    U-net: Con- volutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Con- volutional networks for biomedical image segmentation,” in Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18, pp. 234–241, Springer, 2015. 1, 6

    work page 2015

  8. [8]

    Extraction of olive crown based on uav visible images and the u2-net deep learning model,

    Z. Ye, J. Wei, Y . Lin, Q. Guo, J. Zhang, H. Zhang, H. Deng, and K. Yang, “Extraction of olive crown based on uav visible images and the u2-net deep learning model,” Remote Sens- ing, vol. 14, no. 6, p. 1523, 2022. 1

    work page 2022

Show all 44 references
  1. [9]

    Segmenting purple rapeseed leaves in the field from uav rgb imagery using deep learning as an auxiliary means for nitrogen stress detection,

    J. Zhang, T. Xie, C. Yang, H. Song, Z. Jiang, G. Zhou, D. Zhang, H. Feng, and J. Xie, “Segmenting purple rapeseed leaves in the field from uav rgb imagery using deep learning as an auxiliary means for nitrogen stress detection,” Remote Sensing, vol. 12, no. 9, p. 1403, 2020. 1

    2020

  2. [10]

    Using the u-net convolutional network to map forest types and disturbance in the atlantic rainforest with very high resolution images,

    F. H. Wagner, A. Sanchez, Y . Tarabalka, R. G. Lotte, M. P. Ferreira, M. P. Aidar, E. Gloor, O. L. Phillips, and L. E. Ara- gao, “Using the u-net convolutional network to map forest types and disturbance in the atlantic rainforest with very high resolution images,” Remote Sens...

    2019

  3. [11]

    Mapping forest tree species in high resolution uav-based rgb-imagery by means of convolu- tional neural networks,

    F. Schiefer, T. Kattenborn, A. Frick, J. Frey, P. Schall, B. Koch, and S. Schmidtlein, “Mapping forest tree species in high resolution uav-based rgb-imagery by means of convolu- tional neural networks,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 170, pp. 205–215, 2020. 1

    2020

  4. [12]

    Self-supervised vision transformers for land-cover segmen- tation and classification,

    L. Scheibenreif, J. Hanna, M. Mommert, and D. Borth, “Self-supervised vision transformers for land-cover segmen- tation and classification,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 1422–1431, 2022. 1

    2022

  5. [13]

    A novel multitask transformer deep learning archi- tecture for joint classification and segmentation of horti- culture plantations using very high-resolution satellite im- agery,

    P. Vinod, M. Behera, A. J. Prakash, R. Hebbar, and S. Sri- vastav, “A novel multitask transformer deep learning archi- tecture for joint classification and segmentation of horti- culture plantations using very high-resolution satellite im- agery,” Computers and Electronics in ...

    2024

  6. [14]

    An efficient deep learning mechanism for the recognition of olive trees in jouf region,

    H. H. Alshammari and O. R. Shahin, “An efficient deep learning mechanism for the recognition of olive trees in jouf region,” Computational Intelligence and Neuroscience, vol. 2022, no. 1, p. 9249530, 2022. 1

    2022

  7. [15]

    Spatial pyramid pool- ing in deep convolutional networks for visual recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Spatial pyramid pool- ing in deep convolutional networks for visual recognition,” IEEE transactions on pattern analysis and machine intelli- gence, vol. 37, no. 9, pp. 1904–1916, 2015. 1

    1904

  8. [16]

    Pyramid scene parsing network,

    H. Zhao, J. Shi, X. Qi, X. Wang, and J. Jia, “Pyramid scene parsing network,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2881–2890,

  9. [17]

    Neocognitron: A self- organizing neural network model for a mechanism of visual pattern recognition,

    K. Fukushima and S. Miyake, “Neocognitron: A self- organizing neural network model for a mechanism of visual pattern recognition,” inCompetition and cooperation in neu- ral nets, pp. 267–285, Springer, 1982. 2

    1982

  10. [18]

    Imagenet-trained cnns are biased to- wards texture; increasing shape bias improves accuracy and robustness,

    R. Geirhos, P. Rubisch, C. Michaelis, M. Bethge, F. A. Wich- mann, and W. Brendel, “Imagenet-trained cnns are biased to- wards texture; increasing shape bias improves accuracy and robustness,” arXiv preprint arXiv:1811.12231, 2018. 2

    2018 arXiv

  11. [19]

    Seasonal domain shift in the global south: Dataset and deep features analysis,

    G. V oulgaris, A. Philippides, J. Dolley, J. Reffin, F. Marshall, and N. Quadrianto, “Seasonal domain shift in the global south: Dataset and deep features analysis,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pp. 2115–2123, 2023. 2

    2023

  12. [20]

    Biological con- volutions improve dnn robustness to noise and generalisa- tion,

    B. D. Evans, G. Malhotra, and J. S. Bowers, “Biological con- volutions improve dnn robustness to noise and generalisa- tion,” Neural Networks, vol. 148, pp. 96–110, 2022. 2

    2022

  13. [21]

    Theory of communication. part 1: The analy- sis of information,

    D. Gabor, “Theory of communication. part 1: The analy- sis of information,” Journal of the Institution of Electrical Engineers-Part III: Radio and Communication Engineering, vol. 93, no. 26, pp. 429–441, 1946. 2

    1946

  14. [22]

    Uncertainty relation for resolution in space, spatial frequency, and orientation optimized by two- dimensional visual cortical filters,

    J. G. Daugman, “Uncertainty relation for resolution in space, spatial frequency, and orientation optimized by two- dimensional visual cortical filters,” JOSA A , vol. 2, no. 7, pp. 1160–1169, 1985. 2

    1985

  15. [23]

    Texture image retrieval based on log-gabor features,

    R. Nava, B. Escalante-Ram ´ırez, and G. Crist ´obal, “Texture image retrieval based on log-gabor features,” in Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications: 17th Iberoamerican Congress, CIARP 2012, Buenos Aires, Argentina, September 3-6, 2...

    2012

  16. [24]

    On the choice of band-pass quadrature filters,

    D. Boukerroui, J. A. Noble, and M. Brady, “On the choice of band-pass quadrature filters,”Journal of Mathematical Imag- ing and Vision, vol. 21, pp. 53–80, 2004. 2

    2004

  17. [25]

    Simulation of neural contour mechanisms: from simple to end-stopped cells,

    F. Heitger, L. Rosenthaler, R. V on Der Heydt, E. Peterhans, and O. K ¨ubler, “Simulation of neural contour mechanisms: from simple to end-stopped cells,” Vision research, vol. 32, no. 5, pp. 963–981, 1992. 2

    1992

  18. [26]

    On idempotence and related requirements in edge detection,

    C. Ronse, “On idempotence and related requirements in edge detection,” IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, vol. 15, no. 5, pp. 484–491, 1993. 2

    1993

  19. [27]

    G. H. Granlund and H. Knutsson, Signal processing for com- puter vision. Springer Science & Business Media, 2013. 2

    2013

  20. [28]

    Self-invertible 2d log-gabor wavelets,

    S. Fischer, F. ˇSroubek, L. Perrinet, R. Redondo, and G. Crist ´obal, “Self-invertible 2d log-gabor wavelets,” Inter- national Journal of Computer Vision, vol. 75, pp. 231–246,

  21. [29]

    Gabor con- volutional networks,

    S. Luan, C. Chen, B. Zhang, J. Han, and J. Liu, “Gabor con- volutional networks,” IEEE Transactions on Image Process- ing, vol. 27, no. 9, pp. 4357–4366, 2018. 2

    2018

  22. [30]

    Gabornet: Gabor filters with learnable parameters in deep convolutional neural network,

    A. Alekseev and A. Bobe, “Gabornet: Gabor filters with learnable parameters in deep convolutional neural network,” in International Conference on Engineering and Telecommu- nication, pp. 1–4, 2019. 2

    2019

  23. [31]

    Road extraction by deep residual u-net,

    Z. Zhang, Q. Liu, and Y . Wang, “Road extraction by deep residual u-net,” IEEE Geoscience and Remote Sensing Let- ters, vol. 15, no. 5, pp. 749–753, 2018. 3, 6

    2018

  24. [32]

    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 confer- ence on computer vision and pattern recognition , pp. 770– 778, 2016. 3

    2016

  25. [33]

    Semantic image segmentation with deep con- volutional nets and fully connected crfs,

    L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, “Semantic image segmentation with deep con- volutional nets and fully connected crfs,” arXiv preprint arXiv:1412.7062, 2014. 3

    2014 arXiv

  26. [34]

    Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully con- nected crfs,

    L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, “Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully con- nected crfs,” IEEE transactions on pattern analysis and ma- chine intelligence, vol. 40, no. 4, pp. 834–8...

    2017

  27. [35]

    Re- visiting dilated convolution: A simple approach for weakly- and semi-supervised semantic segmentation,

    Y . Wei, H. Xiao, H. Shi, Z. Jie, J. Feng, and T. S. Huang, “Re- visiting dilated convolution: A simple approach for weakly- and semi-supervised semantic segmentation,” in Proceed- ings of the IEEE conference on computer vision and pattern recognition, pp. 7268–7277, 2018. 3, 4

    2018

  28. [36]

    An im- age is worth 16x16 words: Transformers for image recogni- tion at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An im- age is worth 16x16 words: Transformers for image recogni- tion at scale,” ICLR, 2021. 4

    2021

  29. [37]

    Influence of temperate forest autumn leaf phenology on segmentation of tree species from uav imagery using deep learning,

    M. Cloutier, M. Germain, and E. Lalibert ´e, “Influence of temperate forest autumn leaf phenology on segmentation of tree species from uav imagery using deep learning,” Remote Sensing of Environment, vol. 311, p. 114283, 2024. 5

    2024

  30. [38]

    Landcover.ai: Dataset for automatic mapping of buildings, woodlands, water and roads from aerial imagery,

    A. Boguszewski, D. Batorski, N. Ziemba-Jankowska, T. Dziedzic, and A. Zambrzycka, “Landcover.ai: Dataset for automatic mapping of buildings, woodlands, water and roads from aerial imagery,” in Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition ...

    2021

  31. [39]

    Uavid: A semantic segmentation dataset for uav im- agery,

    Y . Lyu, G. V osselman, G.-S. Xia, A. Yilmaz, and M. Y . Yang, “Uavid: A semantic segmentation dataset for uav im- agery,”ISPRS Journal of Photogrammetry and Remote Sens- ing, vol. 165, pp. 108–119, 2020. 5

    2020

  32. [40]

    Learning deep features for discriminative localization,

    B. Zhou, A. Khosla, A. Lapedriza, A. Oliva, and A. Torralba, “Learning deep features for discriminative localization,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2921–2929, 2016. 6

    2016

  33. [41]

    Multiattention network for semantic seg- mentation of fine-resolution remote sensing images,

    R. Li, S. Zheng, C. Zhang, C. Duan, J. Su, L. Wang, and P. M. Atkinson, “Multiattention network for semantic seg- mentation of fine-resolution remote sensing images,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–13, 2021. 6

    2021

  34. [42]

    Encoder-decoder with atrous separable convolution for se- mantic image segmentation,

    L.-C. Chen, Y . Zhu, G. Papandreou, F. Schroff, and H. Adam, “Encoder-decoder with atrous separable convolution for se- mantic image segmentation,” inProceedings of the European conference on computer vision (ECCV), pp. 801–818, 2018. 6

    2018

  35. [43]

    Unetr: Transformers for 3d medical image segmentation,

    A. Hatamizadeh, Y . Tang, V . Nath, D. Yang, A. Myronenko, B. Landman, H. R. Roth, and D. Xu, “Unetr: Transformers for 3d medical image segmentation,” in Proceedings of the IEEE/CVF winter conference on applications of computer vision, pp. 574–584, 2022. 6

    2022

  36. [44]

    Swin-unet: Unet-like pure transformer for med- ical image segmentation,

    H. Cao, Y . Wang, J. Chen, D. Jiang, X. Zhang, Q. Tian, and M. Wang, “Swin-unet: Unet-like pure transformer for med- ical image segmentation,” in European conference on com- puter vision, pp. 205–218, Springer, 2022. 6

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.