Pith. sign in

REVIEW 4 major objections 5 minor 30 references

OneNet: A Channel-Wise 1D Convolutional U-Net

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read OneNet swaps 2D convolutions for 1D channel-wise convolutions, shrinking U-Net by 47% while keeping segmentation accuracy.

desk verdict Genuinely new architecture combo with real parameter savings, but the accuracy-preservation claim is contradicted by the paper's own Table 1 and the math is too garbled to verify. read the letter →

arxiv 2411.09838 v1 pith:TJZDM3GL submitted 2024-11-14 eess.IV cs.CV

Sanghyun Byun , Kayvan Shah , Ayushi Gang , Christopher Apton , Jacob Song , Woo Seong Chung This is my paper · ORCID
classification eess.IVcs.CV
keywords semanticsegmentationU-Net1Dconvolutionpixel-unshufflepixelshufflemodelcompressionedgedeploymentchannel-wise
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 proposes OneNet, a U-Net variant that replaces 2D convolutions with channel-wise 1D convolutions, using pixel-unshuffle downscaling in the encoder and pixel-shuffle upscaling in the decoder to move spatial information into the channel axis. The central claim is that this design matches the segmentation accuracy of a standard U-Net while cutting parameters by up to 47% with only the encoder swapped, and by up to 71% in model size when the decoder is also replaced. The authors benchmark against U-Net, ResNet, and MobileNet backbones on six mask-prediction datasets, reporting accuracy drops within about 1% on medical tumor segmentation and larger drops on multi-class datasets such as PASCAL VOC and Oxford Pet. If correct, OneNet offers a drop-in efficient backbone for edge deployment, and the pixel-shuffle mechanism becomes a general template for replacing 2D spatial convolutions with cheaper 1D channel operations.

What carries the argument

The load-bearing mechanism is the pixel-unshuffle downscaling operation, which reorganizes a tensor so that each spatial neighborhood of size $s \times s$ becomes a set of extra channels, followed by channel-wise 1D convolutions of kernel size and stride $C_{\text{in}}$ over the flattened $(B, HW C)$ tensor. Because the unshuffle moves spatial relationships into the channel axis, a 1D convolution along that axis can, in the authors' argument, capture the same information a 2D convolution would, at a fraction of the parameters. The decoder mirrors this with pixel-shuffle upscaling that moves information back from channels to space, and an optional spatial 1D convolution with a small kernel is included but shown by ablation to be nearly irrelevant. The paper also implements a 1D-compatible pixel-unshuffle algorithm so that all tensors can be kept in a single spatial dimension.

What would settle it

Take the OneNet encoder and replace the pixel-unshuffle permutation with a fixed random shuffle of the same channel geometry, keeping all other settings unchanged; if accuracy on a medical segmentation dataset does not drop, then the specific spatial organization claimed to carry spatial information is not what is doing the work. Alternatively, if a max-pooling baseline with the same channel counts matches OneNet's accuracy, the unshuffle's information-preservation benefit is not demonstrated.

Watch

Extended reading notes

Core claim

OneNet's central discovery is that a semantic segmentation encoder does not need 2D convolutions if pixel-unshuffle first repackages spatial neighborhoods into the channel dimension. After this repackaging, a channel-wise 1D convolution over the flattened tensor can mix spatial and channel information with a parameter count proportional to $C_{\text{in}} C_{\text{out}}$ rather than $k^2 C_{\text{in}} C_{\text{out}}$, and the receptive field is set by the unshuffle scale rather than by kernel size. In the paper's own formulation, replacing a 2D block with a pixel-unshuffle plus 1D block reduces the per-block multiplication ratio by a factor of $k^2/2$, which translates to 47% fewer parameters for the encoder and a 71% smaller model for the fully 1D version. Across the six datasets, the paper reports that the encoder-only model stays within about 1% of U-Net's accuracy on medical tumor segmentation, while the fully 1D encoder-decoder trades a larger accuracy drop for the largest size reduction.

Load-bearing premise

The whole design rests on the idea that pixel-unshuffle moves spatial relationships into the channel axis, so a channel-wise 1D convolution can see them; if that repackaging does not actually preserve enough spatial information, the encoder loses detail the decoder cannot recover.

Editorial extensions

If this is right

  • OneNet's encoder can replace a standard U-Net encoder in existing pipelines, cutting parameters by roughly 47% while keeping medical-segmentation accuracy within about 1%.
  • A fully 1D encoder-decoder OneNet reaches a 71% smaller model, at the cost of larger accuracy losses on multi-class segmentation datasets.
  • Because the receptive field is controlled by the unshuffle scale rather than kernel size, model designers can tune for local feature detection without changing kernel dimensions.
  • The same pixel-unshuffle-plus-1D block pattern could be applied to other convolutional architectures beyond segmentation, including super-resolution and generative backbones.
  • The ablation suggests spatial convolutions contribute little in this design, so further size reductions are possible by removing them.

Reading between the lines

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

  • A natural extension is to test OneNet as a backbone inside diffusion models, where U-Net is often the computational bottleneck; the authors position the work as adaptable but do not run that experiment.
  • The larger accuracy drops on PASCAL VOC and Oxford Pet suggest the channel-wise 1D design may be better suited to local, low-class-count masks than to image-wide classification; a targeted experiment with high-resolution scenes containing large objects would clarify this.
  • The 1D-compatible pixel-unshuffle algorithm opens the door to using larger unshuffle scales or non-2x2 rearrangements to trade channel capacity for spatial context, which the paper does not explore.
  • One could test whether pretraining the OneNet encoder on ImageNet closes the gap to pretrained ResNet backbones; the paper deliberately does not pretrain, so this remains open.
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 / 5 minor

Summary. The paper proposes OneNet, a U-Net variant that replaces the standard 2D convolutions in the encoder (or encoder-decoder) with channel-wise 1D convolutions combined with pixel-unshuffle and pixel-shuffle operations. The central claim is that this architecture preserves segmentation accuracy while reducing parameters by up to 47% (encoder-only) or 71% (fully 1D encoder-decoder). The method is evaluated on PASCAL VOC, Oxford Pet with two mask versions, and three Medical Segmentation Decathlon datasets (Heart, Brain, Lung), reporting cross-entropy loss, mAP0.5, mIOU, and Dice scores. The paper also reports parameter counts, FLOPs, and memory usage, and includes an ablation on adding spatial convolutions.

Significance. If the central claim were true, OneNet would be a practically useful edge-deployable segmentation backbone, and the combination of pixel-unshuffle with channel-wise 1D convolutions would be an interesting architectural contribution. The paper has some strengths: the code is released, the evaluation covers six datasets, and the ablation on spatial convolutions is a relevant control. However, the main empirical claim is not supported by the paper's own results: Table 1 shows accuracy drops far larger than the 'within 1%' claimed in Section 4.1. In addition, the theoretical derivation in Section 3.1 is built on a garbled equation and an inconsistent algorithm, and the FLOPs comparison in Table 2 uses different input resolutions for the baselines and for OneNet. The architectural premise that pixel-unshuffle moves spatial information into the channel dimension is plausible but not formally justified, and the paper's own measurements do not demonstrate accuracy preservation on general segmentation tasks. The significance of the contribution, as submitted, is therefore not established.

major comments (4)
  1. [Section 4.1, Table 1] The claim 'OneNet achieves model size reduction while maintaining an accuracy drop within 1% across datasets' is directly contradicted by the paper's own numbers. For OneNete,4 against the U-Net4 baseline, mIOU drops from 0.182 to 0.160 on PASCAL VOC (about 12% relative), from 0.316 to 0.216 on Oxford Pet full-mask (about 32% relative), and from 0.713 to 0.636 on the 3-class Pet version (about 11% relative). The encoder-decoder variant OneNeted,4 performs worse still, e.g., 0.149 vs 0.182 on VOC and 0.535 vs 0.713 on Pet-small. The same section later acknowledges 'an 11% and 15% drop in accuracy' and 'a 10% decrease', which is inconsistent with the unqualified 'within 1%' statement. Since the abstract's central claim is that OneNet 'preserves accuracy effectively', this is a load-bearing empirical contradiction that cannot be fixed by rewriting a single sentence.
  2. [Section 4.2, Table 2] The FLOPs and efficiency comparison is invalid because the baselines and OneNet are evaluated at different input resolutions. The text states that a sample tensor of size (1,3,256,256) is used for baselines and (1,3,512,512) for OneNet, while the Table 2 caption says a tensor of size (1,3,256,256) was used as the network input. With OneNet receiving a 512x512 input and the baselines a 256x256 input, the reported FLOPs values are not comparable, and the claimed FLOPs reductions (25% for the encoder, 78% for the full encoder-decoder) are not supported. The parameter-count reduction is a separate quantity, but the computational-efficiency claim in the abstract and Section 4.2 relies on this flawed comparison.
  3. [Section 3.1, Eq. (1), Algorithm 1] The formal definition and the implementation of the proposed pixel-unshuffle downscaling are not internally consistent. Equation (1), D(X)_{i,a,b} = X_{⌊i/s2⌋,s a+⌊sa/i⌋,s b+sb(mod i)}, uses undefined quantities: s2 appears as a divisor but s is already defined as scale, ⌊sa/i⌋ divides by the channel index i, and the term sb(mod i) is not meaningful as written. Algorithm 1 declares an input of shape (B, C, H×W) and an output of shape (B, H×W×C), but the indexing loop and the reshape/transpose/flatten sequence do not produce that output: transpose(0,2,1,3).flatten(dim=1) on a (B,C, HW/4, 4) tensor yields shape (B, HW/4, 4C), not (B, H×W×C). The parameter-count derivation in Section 3.1 depends on this operation, so the theoretical justification for the efficiency gains is not presently verifiable.
  4. [Section 3.2, Section 3.4, Section 4.3] The load-bearing architectural premise is that pixel-unshuffle transfers spatial knowledge to the channel axis, so that channel-wise 1D convolutions on the flattened tensor can replace 2D convolutions. This is asserted in Section 3.2 and again in Section 3.4 ('we can make a mild assumption that the channel dimension already consists of sufficient spatial information'), but it is not demonstrated formally or empirically. The ablation in Section 4.3 only shows that, under this specific design, adding a spatial convolution with kernel size 9 changes accuracy by at most about 2%; it does not show that the channel-wise 1D convolutions actually capture spatial relationships. The accuracy drops on VOC and Oxford Pet are consistent with the alternative hypothesis that the channel-wise 1D encoder loses spatial detail. A correct treatment would need either a formal argument about the information content of the unshuffled tensor or a comparison with a 2D-convolution control that uses the same number of channels, the same downsampling strategy, and the same training protocol.
minor comments (5)
  1. [Section 4.3] The two paragraphs in Section 4.3 are near-duplicates; the second paragraph repeats the description of the ablation and the same reported finding, which makes the section read as if it was accidentally pasted twice.
  2. [Table 1] The shorthand PETF and PETS should be introduced in the caption more explicitly, since the text later refers to 'full-size Oxford Pet' and 'Oxford Pet dataset with fewer masks', which is easy to confuse with the two mask versions.
  3. [Section 3.1, Eq. (1)] Equation (1) should be rewritten with proper index notation and with clearly defined variables; as printed, the subscript ⌊sa/i⌋ and the term sb(mod i) cannot be parsed as a tensor-indexing operation.
  4. [Section 3.2] The sentence 'Receptiveness = S^L' uses a garbled superscript and would be clearer as 'receptive field scales as s^L, where s is the pixel-unshuffle scale and L is the number of layers'.
  5. [Section 4.1] The paper reports LCE (cross-entropy loss) in Table 1 but never analyzes it; either the column should be removed or its relevance to the accuracy claims should be discussed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation is present; the central accuracy claim is empirically contradicted by Table 1, which is a correctness concern, not a circularity.

full rationale

The parameter-reduction claims are derived from explicit arithmetic in Sec. 3.1 and Sec. 3.3, where the paper counts multiplications for 2D versus 1D channel-wise blocks and obtains a k^2/2 ratio; these counts do not use the reported benchmark results, so they are not fitted inputs renamed as predictions. The spatial-information premise is asserted as a design assumption ('we can make a mild assumption that the channel dimension already consists of sufficient spatial information'), not derived from the target accuracy numbers, so it is an untested hypothesis rather than a self-definitional step. No metric in Tables 1 and 3 is constructed from the architecture's definitions; all are independent evaluation outcomes. The only self-citation, reference [3] (MultiDepth), appears in the introduction as an example of a U-Net-based architecture and is not load-bearing for any claim in the derivation. The paper's own Table 1 shows OneNete,4 mIOU dropping from 0.182 to 0.160 on PASCAL VOC and from 0.316 to 0.216 on Oxford Pet full-mask, contradicting the Sec. 4.1 statement of an accuracy drop within 1%; the paper later acknowledges an 11% and 15% drop. That inconsistency concerns empirical correctness and honest reporting, not circularity. Since the only self-citation is non-load-bearing, it does not raise the circularity score under the hard rules.

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

The central efficiency and accuracy claims rest on hand-set architectural choices (scale factor, channel width) and on the unproven premise that pixel-unshuffle transfers spatial information to channels so that channel-wise 1D convolutions suffice. No new physical entities are introduced.

free parameters (4)
  • Pixel-unshuffle scale factor s = 2
    Hand-set to 2 in all experiments; determines channel expansion to 4C and the receptive field S^L, and is central to the parameter-count and accuracy trade-off.
  • Initial bottleneck channel count = 64
    Hand-set base width of the network; directly controls total parameter count and model capacity.
  • Spatial convolution kernel size = 9
    Used only in the ablation study; chosen without a stated search procedure and shown to have negligible effect on accuracy.
  • Background mask loss weight = 1/4
    Hand-set weighting to reduce the background class influence in the cross-entropy loss; affects the reported segmentation metrics.
assumptions (4)
  • standard math Pixel-unshuffle is an information-preserving rearrangement of pixels.
    Used throughout Sec 3.1; true as a permutation when implemented correctly, but Eq. (1) does not actually define it.
  • domain assumption Channel-wise 1D convolution on the pixel-unshuffled tensor captures spatial-channel dependencies.
    Core design premise in Sec 3.2; asserted, not proven, and the paper's own ablation shows spatial convolutions add little without a theoretical explanation.
  • domain assumption The decoder can rely on the encoder's channel dimension containing sufficient spatial information.
    Stated in Sec 3.4 as 'a mild assumption'; this restricts the OneNet decoder to be used only with the OneNet encoder.
  • domain assumption Receptive field equals S^L after L pixel-unshuffle downsampling layers.
    Asserted in Sec 3.2 with no formal analysis of how receptive field accumulates without spatial convolutions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of OneNet: A Channel-Wise 1D Convolutional U-Net." pith.science (2026). https://pith.science/paper/TJZDM3GL

@misc{pith2026241109838,
  author       = {Pith},
  title        = {Pith review of: OneNet: A Channel-Wise 1D Convolutional U-Net},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TJZDM3GL}},
  note         = {Machine review of arXiv:2411.09838}
}
read the original abstract

Many state-of-the-art computer vision architectures leverage U-Net for its adaptability and efficient feature extraction. However, the multi-resolution convolutional design often leads to significant computational demands, limiting deployment on edge devices. We present a streamlined alternative: a 1D convolutional encoder that retains accuracy while enhancing its suitability for edge applications. Our novel encoder architecture achieves semantic segmentation through channel-wise 1D convolutions combined with pixel-unshuffle operations. By incorporating PixelShuffle, known for improving accuracy in super-resolution tasks while reducing computational load, OneNet captures spatial relationships without requiring 2D convolutions, reducing parameters by up to 47%. Additionally, we explore a fully 1D encoder-decoder that achieves a 71% reduction in size, albeit with some accuracy loss. We benchmark our approach against U-Net variants across diverse mask-generation tasks, demonstrating that it preserves accuracy effectively. Although focused on image segmentation, this architecture is adaptable to other convolutional applications. Code for the project is available at https://github.com/shbyun080/OneNet .

Figures

Figures reproduced from arXiv: 2411.09838 by the authors.

Figure 1
Figure 1. Channel-Wise 1D Convolution Block (a) Encoder con￾volution block with pixel-unshuffle downscaling replacing max pooling operation, followed by a single spatial and two channel￾wise layers. (b) Decoder convolution block with pixel-shuffle up￾scaling for tensor upsampling, followed by a spatial layer between two channel-wise layers. model more suitable for lightweight deployment. Addi￾tionally, we optimize spatial pro… view at source ↗
Figure 2
Figure 2. Channel-Wise 1D Encoder-Decoder OneNet employs a U-Net [23] architecture with skip connections for segmentation tasks. The architecture above is a 3-layer variant shown for simplicity. The encoder block replaces the max pool layer with pixel-unshuffle downscaling, with the image downscaled immediately on input for spatial relations to be captured. The decoder block replaces upsampling methods with a pixel-shuffle up… view at source ↗
Figure 3
Figure 3. Comparison of Convolutional Block (a) Traditional 2D convolutional block with max pooling. (b) MobileNet [12] block with max pooling. (c) OneNet implementation with pixel￾unshuffle downscaling followed by 1D convolution. channel on row dimension, as shown below X =  a00 a01 a10 a11 (2) With the traditional 2D approach where the convolu￾tional layer has a kernel size of k = (2, 2) and stride of s = 1 to produce X′ … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 19 canonical work pages

  1. [1]

    Land- man, Geert Litjens, Bjoern Menze, Olaf Ronneberger, Ronald M

    Michela Antonelli, Annika Reinke, Spyridon Bakas, Key- van Farahani, Annette Kopp-Schneider, Bennett A. Land- man, Geert Litjens, Bjoern Menze, Olaf Ronneberger, Ronald M. Summers, Bram van Ginneken, Johanna Kirch- berg, Fabian Isensee, Klaus H. Maier-Hein, M. Jorge Car- doso, Ruben Janssens, Paul F. J ¨ager, Simon Kohl, Laura Lange, Suprosanna Shit, Chri...

    arXiv 2022

  2. [2]

    Blitzmask: Real-time instance segmentation approach for mobile devices

    Vitalii Bulygin, Dmytro Mykheievskyi, and Kyrylo Kuchyn- skyi. Blitzmask: Real-time instance segmentation approach for mobile devices. In Proceedings of The 26th International Conference on Artificial Intelligence and Statistics , pages 1799–1811. PMLR, 2023. 2

    work page 2023

  3. [3]

    MultiDepth: Multi-Sample Priors for Refining Monocular Metric Depth Estimations in Indoor Scenes

    Sanghyun Byun, Jacob Song, and Woo Seong Chung. Mul- tidepth: Multi-sample priors for refining monocular metric depth estimations in indoor scenes.arXiv:2411.01048, 2024. 1

    work page Pith review arXiv 2024

  4. [4]

    N ¨urnberger

    Soumick Chatterjee, Alessandro Sciarra, Max D ¨unnwald, Raghava Vinaykanth Mushunuri, Ranadheer Podishetti, Ra- jatha Nagaraja Rao, Geetha Doddapaneni Gopinath, Stef- fen Oeltze-Jafra, Oliver Speck, and A. N ¨urnberger. Shuffle- unet: Super resolution of diffusion-weighted mris using deep learning. 2021 29th European Signal Processing Conference (EUSIPCO)...

    work page 2021

  5. [5]

    Zheng Chen, Zongwei Wu, Eduard Zamfir, Kai Zhang, Yu- lun Zhang, Radu Timofte, Xiaokang Yang, Hongyuan Yu, Cheng Wan, Yuxin Hong, Zhijuan Huang, Yajun Zou, Yuan Huang, Jiamin Lin, Bingnan Han, Xianyu Guan, Yong- sheng Yu, Daoan Zhang, Xuanwu Yin, Kunlong Zuo, Jin- hua Hao, Kai Zhao, Kun Yuan, Ming Sun, Chao Zhou, Hongyu An, Xinfeng Zhang, Zhiyuan Song, Zi...

    work page 2024

  6. [6]

    Deep neural networks segment neu- ronal membranes in electron microscopy images

    Dan Ciresan, Alessandro Giusti, Luca Gambardella, and J¨urgen Schmidhuber. Deep neural networks segment neu- ronal membranes in electron microscopy images. In Ad- vances in Neural Information Processing Systems . Curran Associates, Inc., 2012. 2

    work page 2012

  7. [7]

    Soham De and Samuel L. Smith. Batch normalization bi- ases residual blocks towards the identity function in deep networks. In Proceedings of the 34th International Confer- ence on Neural Information Processing Systems, Red Hook, NY , USA, 2020. Curran Associates Inc. 2

    work page 2020

  8. [8]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition, pages 248–255. IEEE, 2009. 6

    work page 2009

Show all 30 references
  1. [9]

    The pascal visual object classes (voc) challenge

    Mark Everingham, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. The pascal visual object classes (voc) challenge. International Journal of Computer Vision, 88(2):303–338, 2010. 5, 6

    2010

  2. [10]

    Brain tumor segmentation from mri images using deep learning techniques

    Ayan Gupta, Mayank Dixit, Vipul Kumar Mishra, Attulya Singh, and Atul Dayal. Brain tumor segmentation from mri images using deep learning techniques. arXiv:2305.00257,

    arXiv

  3. [11]

    Zhang, Shaoqing Ren, and Jian Sun

    Kaiming He, X. Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. 2016 IEEE Confer- ence on Computer Vision and Pattern Recognition (CVPR) , pages 770–778, 2015. 1, 2, 5, 6, 7

    2016

  4. [12]

    Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco An- dreetto, and Hartwig Adam

    Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco An- dreetto, and Hartwig Adam. Mobilenets: Efficient convolu- tional neural networks for mobile vision applications.CoRR, abs/1704.04861, 2017. 1, 2, 4, 5, 6, 7

    2017 arXiv

  5. [13]

    Adam: A method for stochastic optimization

    Diederik Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In International Conference on Learning Representations (ICLR), 2015. 6

    2015

  6. [14]

    Convolutional net- works with oriented 1d kernels

    Alexandre Kirchmeyer and Jia Deng. Convolutional net- works with oriented 1d kernels. arXiv:2309.15812, 2023. 2

    2023 arXiv

  7. [15]

    A 1d convolutional network for leaf and time series classification

    Dongyang Kuang. A 1d convolutional network for leaf and time series classification. arXiv:1907.00069, 2020. 2

    1907 arXiv

  8. [16]

    Tfnet: Exploiting temporal cues for fast and accurate lidar semantic segmentation

    Rong Li, ShiJie Li, Xieyuanli Chen, Teli Ma, Juergen Gall, and Junwei Liang. Tfnet: Exploiting temporal cues for fast and accurate lidar semantic segmentation. arXiv:2309.07849, 2024. 2, 3

    2024 arXiv

  9. [17]

    Mo- bilevig: Graph-based sparse attention for mobile vision ap- plications

    Mustafa Munir, William Avery, and Radu Marculescu. Mo- bilevig: Graph-based sparse attention for mobile vision ap- plications. arXiv:2307.00395, 2023. 2, 3

    2023 arXiv

  10. [18]

    Eunnet: Efficient un-normalized convolution layer for stable training of deep residual networks without batch normaliza- tion layer

    Khanh-Binh Nguyen, Jaehyuk Choi, and Joon-Sung Yang. Eunnet: Efficient un-normalized convolution layer for stable training of deep residual networks without batch normaliza- tion layer. IEEE Access, 2023. 2

    2023

  11. [19]

    Ozan Oktay, Jo Schlemper, Lo ¨ıc Le Folgoc, Matthew C. H. Lee, Mattias P. Heinrich, Kazunari Misawa, Kensaku Mori, Steven G. McDonagh, Nils Y . Hammerla, Bernhard Kainz, Ben Glocker, and Daniel Rueckert. Attention u-net: Learn- ing where to look for the pancreas. CoRR, abs/1804.03999,

    arXiv

  12. [20]

    Parkhi, Andrea Vedaldi, Andrew Zisserman, and C

    Omkar M. Parkhi, Andrea Vedaldi, Andrew Zisserman, and C. V . Jawahar. Cats and dogs. 2012 IEEE Conference on Computer Vision and Pattern Recognition , pages 3498– 3505, 2012. 5, 6 8

    2012

  13. [21]

    Automatic differentiation in pytorch

    Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Al- ban Desmaison, Luca Antiga, and Adam Lerer. Automatic differentiation in pytorch. In NIPS-W, 2017. 6

    2017

  14. [22]

    Sam 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman R¨adle, Chloe Rolland, Laura Gustafson, Eric Mintun, Junt- ing Pan, Kalyan Vasudev Alwala, Nicolas Carion, Chao- Yuan Wu, Ross Girshick, Piotr Doll´ar, and Christoph Feic...

    2024 arXiv

  15. [23]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. CoRR, abs/1505.04597, 2015. 1, 2, 3, 4, 5, 6, 7

    2015 arXiv

  16. [24]

    Aitken, Rob Bishop, Daniel Rueckert, and Zehan Wang

    Wenzhe Shi, Jose Caballero, Ferenc Husz ´ar, Johannes Totz, Andrew P. Aitken, Rob Bishop, Daniel Rueckert, and Zehan Wang. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network. CoRR, abs/1609.05158, 2016. 1, 2

    2016 arXiv

  17. [25]

    Freeu: Free lunch in diffusion u-net

    Chenyang Si, Ziqi Huang, Yuming Jiang, and Ziwei Liu. Freeu: Free lunch in diffusion u-net. arXiv:2309.11497,

    arXiv

  18. [26]

    Pp-mobileseg: Explore the fast and accurate semantic segmentation model on mobile devices

    Shiyu Tang, Ting Sun, Juncai Peng, Guowei Chen, Yuy- ing Hao, Manhui Lin, Zhihong Xiao, Jiangbin You, and Yi Liu. Pp-mobileseg: Explore the fast and accurate semantic segmentation model on mobile devices. arXiv:2304.05152,

    arXiv

  19. [27]

    One-dimensional deep low-rank and sparse network for accelerated mri

    Zi Wang, Chen Qian, Di Guo, Hongwei Sun, Rushuai Li, Bo Zhao, and Xiaobo Qu. One-dimensional deep low-rank and sparse network for accelerated mri. IEEE Transactions on Medical Imaging, 42(1):79–90, 2023. 2

    2023

  20. [28]

    Harmonyview: Harmonizing consis- tency and diversity in one-image-to-3d

    Sangmin Woo, Byeongjun Park, Hyojun Go, Jin-Young Kim, and Changick Kim. Harmonyview: Harmonizing consis- tency and diversity in one-image-to-3d. arXiv:2312.15980,

    arXiv

  21. [29]

    Unet++: A nested u- net architecture for medical image segmentation

    Zongwei Zhou, Md Mahfuzur Rahman Siddiquee, Nima Tajbakhsh, and Jianming Liang. Unet++: A nested u- net architecture for medical image segmentation. CoRR, abs/1807.10165, 2018. 2

    2018 arXiv

  22. [30]

    In- fer from what you have seen before: Temporally-dependent classifier for semi-supervised video segmentation

    Jiafan Zhuang, Zilei Wang, Yixin Zhang, and Zhun Fan. In- fer from what you have seen before: Temporally-dependent classifier for semi-supervised video segmentation. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3575–3584,...

    2024

Pith tools

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