Pith. sign in

REVIEW 3 major objections 6 minor 44 references

UNet--: Memory-Efficient and Feature-Enhanced Network Architecture based on U-Net with Reduced Skip-Connections

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Replacing U-Net's four stored skip-connection feature maps with a single compact map cut the memory they demand by 93.3% while improving restoration accuracy.

desk verdict UNet-- is a credible, well-scoped engineering contribution that cuts skip-connection storage by an order of magnitude, but the headline memory claim is only about one memory component and the paper never measures total or peak activation memory. read the letter →

arxiv 2412.18276 v1 pith:FXHMBOTZ submitted 2024-12-24 cs.CV eess.IV

Lingxiao Yin , Wei Tao , Dongyue Zhao , Tadayuki Ito , Kinya Osa , Masami Kato , Tse-Wei Chen This is my paper · ORCID
classification cs.CVeess.IV
keywords U-Netskipconnectionsmemoryefficiencyimagerestorationmulti-scalefeatureaggregationenhancementNAFNetdenoising
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 claims that the memory cost of U-Net's skip-connections can be reduced by 93.3% by replacing the four multi-scale encoder feature maps held in memory with a single compact aggregated map, then regenerating the multi-scale features in the decoder. The regeneration is done by a proposed Information Enhancement Module (IEM) that produces feature maps with higher diversity than the originals, which is why accuracy does not drop—it improves. Applied to NAFNet, the architecture gains 0.04 PSNR on denoising, 0.19 on deblurring, and 0.398 on super-resolution, at 7.9% more MACs and 2.8% more parameters. The method is presented as universal, with experiments also on image matting using a different backbone.

What carries the argument

The central mechanism is the pair MSIAM + IEM. MSIAM reduces each encoder feature map's channels with point-wise convolutions, resizes them to the coarsest resolution (H/8 × W/8) with pixel unshuffles, concatenates them, and fuses them with a point-wise convolution; this produces a single compact map that is the only thing held in memory. IEM then pixel-shuffles that map back to the four target resolutions and passes each through an enhancement head made of a ConvNeXt V2 block and a separable convolution, producing feature maps with the same shapes as the originals but with enhanced information. The stored state becomes the compact map rather than the sum of four full-resolution encoder maps.

What would settle it

Measure the peak activation memory of the full UNet-- during inference on a representative input (e.g., 256×256, 32 channels) including all intermediate decoder feature maps; if the peak is not at least 50% below the baseline NAFNet's peak, the claimed memory reduction is not realized.

Watch

Extended reading notes

Core claim

The central claim is that the multi-scale feature maps U-Net stores for its skip connections can be collapsed into a single-scale, low-memory representation during encoding and then expanded back into richer multi-scale features during decoding, so the network only needs to hold the compact map between the two phases. The paper proposes MSIAM to aggregate the encoder's four feature maps—after channel reduction and resolution alignment—into one map of the smallest resolution, and IEM to resize and enhance that map back into the four original resolutions. The authors report that the memory held for the skip connections drops from 3.75 MB to 0.25 MB for NAFNet-sized inputs, a 93.3% reduction, while PSNR improves on all three restoration tasks; they also report that the IEM-generated features are not imitations of the encoder outputs (low SSIM) and have higher channel diversity (higher variance of pairwise cosine distances).

Load-bearing premise

The memory comparison counts only the four feature maps held for the skip connections; it assumes the IEM-generated multi-scale maps, which are created during decoding and have the same resolutions as the original skip maps, either do not need to be held in memory simultaneously or cost less than the maps they replace.

Editorial extensions

If this is right

  • Deploying U-Net-like restoration models on memory-constrained hardware becomes more practical: the stored activation for skip connections drops from 3.75 MB to 0.25 MB at 256×256 inputs, with total MACs and parameters increasing only modestly.
  • The same MSIAM/IEM swap can be applied to any U-Net variant; the matting experiment on MSCANtiny shows 94.5% skip-memory reduction with improved or comparable quality.
  • Because the IEM feature maps are generated on the fly during decoding, the encoder does not need to hold multiple resolution maps, which changes how U-Net memory is budgeted on accelerators.
  • The gain is not from simply dropping skip connections: the no-skip baselines lose 0.4 PSNR on denoising, while UNet-- gains 0.04, so the enhancement matters rather than just the removal.

Reading between the lines

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

  • If the 93.3% figure is confirmed against total activation memory, the same aggregation-then-regeneration idea could be applied to other multi-scale architectures, such as feature pyramid networks, to compress their stored feature hierarchies.
  • The low SSIM between IEM outputs and encoder outputs suggests the IEM learns a mapping to a different, possibly more decorrelated feature space; this could be tested by measuring whether the downstream decoder's effective receptive field changes.
  • The paper's memory analysis only counts skip-map storage; a fair comparison of peak on-chip memory would require measuring the full activation footprint, including the IEM's intermediate maps.
  • One could test the enhancement claim directly by ablating the ConvNeXt V2 block in the IEM: if PSNR does not drop, the improvement may come from the aggregation rather than the enhancement.
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

3 major / 6 minor

Summary. The paper proposes UNet--, an architecture that replaces the multi-scale skip-connection feature maps of a U-Net with a single compact feature map produced by a Multi-Scale Information Aggregation Module (MSIAM) in the encoder, and regenerates enhanced multi-scale feature maps in the decoder through an Information Enhancement Module (IEM). The method is applied to NAFNet for image denoising, deblurring, and super-resolution, and to MSCANtiny for image matting. The authors report a 93.3% reduction in the memory held for skip connections (Msc), with small PSNR gains on denoising, deblurring, and super-resolution, and claim improvements on matting.

Significance. The core idea—aggregating multi-scale encoder features into a compact representation and regenerating them in the decoder—is a plausible and potentially useful direction for memory-efficient U-Net design, and the module is simple enough to be adopted in other U-Net variants. The reported Msc reductions are consistent with the architecture's design. However, the central 'memory-efficient' claim is currently scoped only to persistent skip-connection buffers, not to total or peak activation memory; the transient generation cost of IEM is never measured, leaving the deployment-oriented claim unverified. In addition, the matting results as reported contradict the stated improvements. If these issues are resolved, the contribution would be of interest to the efficient image restoration community.

major comments (3)
  1. [Sec. 3.1, Fig. 1(b), Tabs. 4-5] The 93.3% memory reduction is measured only for Msc, the persistent feature maps held for the four skip connections (peak 15/8 ME1 = 3.75 MB for NAFNet, versus 0.25 MB for UNet--). The paper does not measure total or peak activation memory for either network. In the decoder, IEM produces feature maps with the same resolutions and channel counts as E1-E4, and the pixel-shuffle operation for restoring the E1-resolution map (stride 8) requires an intermediate 2048-channel 32x32 tensor (2 MB at the reported 256x256 input) plus the 2 MB output, in addition to the persistent 0.25 MB single-scale map. At the D1 stage, the original NAFNet holds only the 2 MB E1 map, so UNet--'s per-stage activation peak can exceed NAFNet's. Please report full-network activation memory traces (or at least peak activation memory during inference) and clarify whether IEM generates all four maps up front or stage-by-stage; without this, the title/abstract claim of memory efficiency is not established for actual deployment.
  2. [Sec. 4.4, Tab. 7] The text states that MSCANtiny with UNet-- 'outperforms the baseline model MSCANtiny with 0.246 (whole image), 0.402 (unknown region) in MSE', but Table 7 shows the opposite: every whole-image and unknown-region MSE entry for UNet-- is worse than the corresponding baseline entry (e.g., whole-image MSE on RWP-636: 14.835 vs 14.107, on P3M-500-NP: 2.391 vs 2.369, on P3M-500-P: 3.042 vs 2.998). The SAD entries are also mostly worse. This contradiction undermines the claim of consistent improvement across tasks. Please correct the text or provide the missing experiments/definitions of the quoted deltas.
  3. [Tabs. 4-6] The reported accuracy gains are small (e.g., +0.04 PSNR and +0.000 SSIM in denoising, Table 4) and no variance or multiple-seed results are provided. Without error bars or repeated runs, these differences may be within run-to-run noise. Please report mean +/- std over at least three training runs for the main comparisons, or otherwise demonstrate that the improvements are reproducible and statistically meaningful.
minor comments (6)
  1. [Sec. 1, Sec. 3.3] There are several typos: 'IEM moduel' should be 'IEM module', 'enhancement header' should be 'enhancement head', and 'unsampling' should be 'upsampling'.
  2. [Sec. 3.3, Fig. 3] The three architecture variants in Fig. 3 are described but not experimentally compared. An ablation of the target resolution (e.g., minimum vs maximum resolution aggregation) would help justify the choice used in the NAFNet experiments.
  3. [Sec. 3.3, Tabs. 2-3] The 'representative ability' metric (variance of pairwise cosine distances between channels) is non-standard, and the paper does not demonstrate a direct relationship between this metric and end-task performance. Please provide a more standard diversity metric or an explicit correlation analysis.
  4. [Tab. 6] The super-resolution results omit Msc, MACs, and Params columns; including them would make the memory and complexity claims complete, especially since the abstract claims consistent memory improvements across tasks.
  5. [Sec. 4.4] The text reports SSIM improvements for matting, but Table 7 lists MSE and SAD; please align the reported metrics.
  6. [General] No code or model release is mentioned, which would aid reproducibility of the reported numbers; consider adding a footnote with the intended release plan.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the memory reduction is arithmetic from the stated design, and the accuracy gains are measured on external benchmarks.

full rationale

The paper's central claims are not circular. The 93.3% reduction in skip-connection memory is computed directly in Sec. 3.1 from the sizes of the feature maps that are held in memory: NAFNet stores {E1,E2,E3,E4} totaling 15/8*ME1 = 3.75 MB, while UNet-- stores the single-scale MSIAM output of 0.25 MB. This is a design-level accounting of persistent skip-connection buffers, not a fitted parameter presented as a prediction. The performance improvements (denoising +0.04 PSNR, deblurring +0.19, super-resolution +0.398) are measured on held-out benchmarks (SIDD, GoPro, Set5/Set14/B100/Urban100/Manga109) against baseline NAFNet, and they are not derived from the memory-saving equations. The similarity and representative-ability analyses in Sec. 3.3 (Tabs. 2 and 3) are empirical comparisons, not definitions that presuppose the conclusion. No load-bearing self-citation appears; the paper does not invoke prior work by its own authors to justify its architecture. The possible concern that total activation memory, including transient IEM-generated maps, may offset the reported savings is an evaluation-scope or correctness risk, not a circularity, because the paper explicitly scopes its claim to 'the memory demand by the skip-connections' and does not hide the existence of the IEM-generated maps. Thus the derivation chain is self-contained: the savings follow by construction from the architecture, and the accuracy claims are externally falsifiable.

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

The central claim rests on a small number of hand-chosen design parameters (channel reduction ratios, target resolution, enhancement block choice) rather than fitted constants. No new physical or mathematical entities are introduced.

free parameters (3)
  • MSIAM channel reduction ratios = {1/16, 1/16, 1/16, 1/8}
    Chosen by hand in Sec. 4 to compress encoder outputs E1-E4 to {2,4,8,32} channels; not tuned per task, but an ad hoc design choice that determines memory savings and information retention.
  • Pixel unshuffle/shuffle strides = {8,4,2} for E1-E3
    Set in Sec. 4 to bring E1-E3 down to the E4 resolution (H/8 x W/8); the target resolution (E4) is a design decision, with variants shown in Fig. 3.
  • IEM enhancement block structure = ConvNeXt V2 block + separable convolution
    Module choice from Sec. 3.3; not derived, selected for representative ability and cost.
assumptions (3)
  • domain assumption Skip-connection feature maps must be held in memory until decoding reaches the matching decoder stage
    Sec. 3.1 memory analysis; standard for U-Net inference, but ignores recomputation or streaming alternatives.
  • domain assumption Aggregating multi-scale encoder features into one map preserves sufficient information to reconstruct enhanced multi-scale decoder features
    Central design premise of MSIAM/IEM in Sec. 3.3; supported only by the empirical results and the SSIM/variance analysis in Tabs. 2-3.
  • domain assumption ConvNeXt V2 block provides strong representative ability at acceptable cost
    Invoked in Sec. 3.3 IEM design; taken from prior work [33].

how reviews work

0 comments
Cite this review

Pith. "Pith review of UNet--: Memory-Efficient and Feature-Enhanced Network Architecture based on U-Net with Reduced Skip-Connections." pith.science (2026). https://pith.science/paper/FXHMBOTZ

@misc{pith2026241218276,
  author       = {Pith},
  title        = {Pith review of: UNet--: Memory-Efficient and Feature-Enhanced Network Architecture based on U-Net with Reduced Skip-Connections},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FXHMBOTZ}},
  note         = {Machine review of arXiv:2412.18276}
}
read the original abstract

U-Net models with encoder, decoder, and skip-connections components have demonstrated effectiveness in a variety of vision tasks. The skip-connections transmit fine-grained information from the encoder to the decoder. It is necessary to maintain the feature maps used by the skip-connections in memory before the decoding stage. Therefore, they are not friendly to devices with limited resource. In this paper, we propose a universal method and architecture to reduce the memory consumption and meanwhile generate enhanced feature maps to improve network performance. To this end, we design a simple but effective Multi-Scale Information Aggregation Module (MSIAM) in the encoder and an Information Enhancement Module (IEM) in the decoder. The MSIAM aggregates multi-scale feature maps into single-scale with less memory. After that, the aggregated feature maps can be expanded and enhanced to multi-scale feature maps by the IEM. By applying the proposed method on NAFNet, a SOTA model in the field of image restoration, we design a memory-efficient and feature-enhanced network architecture, UNet--. The memory demand by the skip-connections in the UNet-- is reduced by 93.3%, while the performance is improved compared to NAFNet. Furthermore, we show that our proposed method can be generalized to multiple visual tasks, with consistent improvements in both memory consumption and network accuracy compared to the existing efficient architectures.

Figures

Figures reproduced from arXiv: 2412.18276 by the authors.

Figure 1
Figure 1. (a) Architecture of a common U-Net model. The dashed lines represent the skip￾connections. Feature maps in the blue shadow are needed to be maintained in memory. (b) Memory consumption for skip-connections for U-Net and UNet−− throughout the inference process, the unit is ME1 which denotes the memory consumption of E1. many tasks e.g., image restoration [5], image segmentation [4,6,10], image mat￾ting [36], etc. U-N… view at source ↗
Figure 2
Figure 2. Illustration of the proposed UNet−− network. The original skip-connections are replaced with MSIAM and IEM. MSIAM aggregates multi-scale feature maps to single-scale with less memory demand. IEM generates enhanced multi-scale feature maps according to the output of MSIAM. Feature maps in the blue shadow are needed to be maintained in memory. skip-connection between encoder and decoder. The four U-Net models share th… view at source ↗
Figure 3
Figure 3. Model structure variants corresponding to different target resolution. Feature maps in the blue shadow are needed to be maintained in memory. (a) The resolution of the aggregated feature maps equals to the minimum resolution in the encoder. (b) The resolution of the aggregated feature maps equals to the intermediate resolution in the encoder. (c) The resolution of the aggregated feature map equals to the maximum res… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Visualization results. The first row includes noisy images, the second row in￾cludes clean images outputted by NAFNet [5] with UNet−−, the third row shows the comparison of local details. cost of skip-connections without causing harm to the model performance in an effi…
Figure 5
Figure 5. Figure 5: Visualization results. The first row includes blurry image, the second row in￾cludes clean images outputted by NAFNet [5] with UNet−−, the third row shows the comparison of local details. The evaluation result is shown in Tab. 5. With the same 93.3% memory savings, the…
Figure 6
Figure 6. Figure 6: Visualization results of the proposed NAFNet [5] with UNet−− on image super resolution. 4.4 Image matting Image matting is a fundamental computer vision task which aims to predict an alpha matte to precisely cut out an image region. It estimates the foreground, backgro…
Figure 7
Figure 7. Figure 7: Visualization results of the proposed MSCANtiny [10] with UNet−− on image matting. of absolute differences and the mean squared error between the ground-truth alpha matte and the predicted alpha matte, respectively. For each model, we calculate the average values of MS…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 28 canonical work pages

  1. [1]

    Abdelhamed, A., Lin, S., Brown, M.S.: A high-quality denoising dataset for smart- phone cameras. In: Proceedings of the IEEE/CVF Conference on Computer Vision andPatternRecognition.pp.1692–1700(April2018),https://www.microsoft.com/ en-us/research/publication/high-quality-denoising-dataset-smartphone-cameras/

    work page

  2. [2]

    In: 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

    Agustsson, E., Timofte, R.: Ntire 2017 challenge on single image super-resolution: Dataset and study. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). pp. 1122–1131 (2017). https://doi.org/10. 1109/CVPRW.2017.150

    work page 2017

  3. [3]

    Bevilacqua, M., Roumy, A., Guillemot, C.M., Alberi-Morel, M.L.: Low-complexity singleimagesuper-resolutionbasedonnonnegativeneighborembedding.In:British Machine Vision Conference (2012)

    work page 2012

  4. [4]

    In: 2019 IEEE/CVF International Conference on Computer Vision (ICCV)

    Bolya, D., Zhou, C., Xiao, F., Lee, Y.J.: Yolact: Real-time instance segmentation. In: 2019 IEEE/CVF International Conference on Computer Vision (ICCV). pp. 9156–9165 (2019). https://doi.org/10.1109/ICCV.2019.00925

    arXiv 2019

  5. [5]

    arXiv preprint arXiv:2204.04676 (2022)

    Chen, L., Chu, X., Zhang, X., Sun, J.: Simple baselines for image restoration. arXiv preprint arXiv:2204.04676 (2022)

    arXiv 2022

  6. [6]

    In: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Cheng, T., Wang, X., Chen, S., Zhang, W., Zhang, Q., Huang, C., Zhang, Z., Liu, W.: Sparse instance activation for real-time instance segmentation. In: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 4423–4432 (2022). https://doi.org/10.1109/CVPR52688.2022.00439

    arXiv 2022

  7. [7]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Ding, X., Zhang, X., Ma, N., Han, J., Ding, G., Sun, J.: Repvgg: Making vgg-style convnets great again. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 13733–13742 (2021)

    work page 2021

  8. [8]

    In: 2022 IEEE/CVF Conference on Com- puter Vision and Pattern Recognition Workshops (CVPRW)

    Du, Z., Liu, D., Liu, J., Tang, J., Wu, G., Fu, L.: Fast and memory-efficient network towards efficient image super-resolution. In: 2022 IEEE/CVF Conference on Com- puter Vision and Pattern Recognition Workshops (CVPRW). pp. 852–861 (2022). https://doi.org/10.1109/CVPRW56347.2022.00101

    arXiv 2022

Show all 44 references
  1. [9]

    IEEE Transactions on Pattern Analysis and Machine Intelligence 43(2), 652–662 (Feb 2021)

    Gao, S.H., Cheng, M.M., Zhao, K., Zhang, X.Y., Yang, M.H., Torr, P.: Res2net: A new multi-scale backbone architecture. IEEE Transactions on Pattern Analysis and Machine Intelligence 43(2), 652–662 (Feb 2021). https://doi.org/10.1109/ tpami.2019.2938758, http://dx.doi.org/10.11...

    2021

  2. [10]

    ArXivabs/2209.08575 (2022), https://api.semanticscholar.org/CorpusID:252367800

    Guo, M.H., Lu, C., Hou, Q., Liu, Z., Cheng, M.M., Hu, S.: Segnext: Rethinking convolutional attention design for semantic segmentation. ArXivabs/2209.08575 (2022), https://api.semanticscholar.org/CorpusID:252367800

    2022 arXiv

  3. [11]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 770–778 (2016)

    2016

  4. [12]

    Huang, H., Lin, L., Tong, R., Hu, H., Zhang, Q., Iwamoto, Y., Han, X., Chen, Y.W., Wu, J.: Unet 3+: A full-scale connected unet for medical image segmentation (2020)

    2020

  5. [13]

    IEEE (2015)

    Huang, J.B., Singh, A., Ahuja, N.: Single image super-resolution from transformed self-exemplars. IEEE (2015)

    2015

  6. [14]

    In: Medical Image Computing and Computer Assisted Intervention

    Ibtehaz, N., Kihara, D.: Acc-unet: A completely convolutional unet model for the 2020s. In: Medical Image Computing and Computer Assisted Intervention. vol. 14222 (2023)

    2023

  7. [15]

    In: 2020 IEEE International Conference on Image Processing (ICIP)

    Li, C., Tan, Y., Chen, W., Luo, X., Gao, Y., Jia, X., Wang, Z.: Attention unet++: A nested attention-aware u-net for liver ct image segmentation. In: 2020 IEEE International Conference on Image Processing (ICIP). pp. 345–349 (2020). https: //doi.org/10.1109/ICIP40778.2020.9190...

    2020

  8. [16]

    In: Pro- ceedings of the 29th ACM International Conference on Multimedia

    Li, J., Ma, S., Zhang, J., Tao, D.: Privacy-preserving portrait matting. In: Pro- ceedings of the 29th ACM International Conference on Multimedia. p. 3501–3509. MM ’21, Association for Computing Machinery, New York, NY, USA (2021). https: //doi.org/10.1145/3474085.3475512, htt...

    2021

  9. [17]

    IEEE Computer Society (2017)

    Lin, T.Y., Dollár, P., Girshick, R., He, K., Hariharan, B., Belongie, S.: Feature pyramid networks for object detection. IEEE Computer Society (2017)

    2017

  10. [18]

    Loshchilov, I., Hutter, F.: Sgdr: Stochastic gradient descent with restarts (08 2016)

    2016

  11. [19]

    IEEE (2002)

    Martin, D., Fowlkes, C., Tal, D., Malik, J.: A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics. IEEE (2002)

    2002

  12. [20]

    Multimedia Tools and Applications (2017)

    Matsui, Y., Ito, K., Aramaki, Y., Fujimoto, A., Ogawa, T., Yamasaki, T., Aizawa, K.: Sketch-based manga retrieval using manga109 dataset. Multimedia Tools and Applications (2017)

    2017

  13. [21]

    In: 2017 IEEE Conference on Computer Vision and PatternRecognition(CVPR).pp.257–265(2017).https://doi.org/10.1109/CVPR

    Nah, S., Kim, T.H., Lee, K.M.: Deep multi-scale convolutional neural network for dynamic scene deblurring. In: 2017 IEEE Conference on Computer Vision and PatternRecognition(CVPR).pp.257–265(2017).https://doi.org/10.1109/CVPR. 2017.35

    2017 doi

  14. [22]

    In: Medical Imaging with Deep Learning (2018)

    Oktay, O., Schlemper, J., Folgoc, L.L., Lee, M., Heinrich, M., Misawa, K., Mori, K., McDonagh, S., Hammerla, N.Y., Kainz, B., Glocker, B., Rueckert, D.: Attention u-net: Learning where to look for the pancreas. In: Medical Imaging with Deep Learning (2018)

    2018

  15. [23]

    In: International Conference on Medical Image Computing and Computer-Assisted Intervention (2015)

    Ronneberger, O., Fischer, P., Brox, T.: U-net: Convolutional networks for biomedi- cal image segmentation. In: International Conference on Medical Image Computing and Computer-Assisted Intervention (2015)

    2015

  16. [24]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Shelhamer, E., Long, J., Darrell, T.: Fully convolutional networks for semantic segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. p. 3431–3440 (2015)

    2015

  17. [25]

    Genome Biology25, 49 (2024)

    Singh, G., Alser, M., Denolf, K., Firtina, C., Khodamoradi, A., Cavlak, M.B., Cor- poraal, H., Mutlu, O.: Rubicon: a framework for designing efficient deep learning- based genomic basecallers. Genome Biology25, 49 (2024)

    2024

  18. [26]

    Sun, G., Pan, Y., Kong, W., Xu, Z., Ma, J., Racharak, T., Nguyen, L.M., Xin, J.: Da-transunet: Integrating spatial and channel dual attention with transformer u-net for medical image segmentation (2023)

    2023

  19. [27]

    Szegedy, C., Ioffe, S., Vanhoucke, V., Alemi, A.: Inception-v4, inception-resnet and the impact of residual connections on learning (2016)

    2016

  20. [28]

    Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., Wojna, Z.: Rethinking the incep- tion architecture for computer vision. IEEE pp. 2818–2826 (2016)

    2016

  21. [29]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (2023)

    Vasu, P.K.A., Gabriel, J., Zhu, J., Tuzel, O., Ranjan, A.: Fastvit: A fast hy- brid vision transformer using structural reparameterization. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (2023)

    2023

  22. [30]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2023)

    Vasu, P.K.A., Gabriel, J., Zhu, J., Tuzel, O., Ranjan, A.: An improved one millisec- ond mobile backbone. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2023)

    2023

  23. [31]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Wang, H., Cao, P., Wang, J., Zaiane, O.R.: Uctransnet: Rethinking the skip con- nections in u-net from a channel-wise perspective with transformer. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 36, pp. 2441–2449 (2022)

    2022

  24. [32]

    ACM Transactions on Reconfigurable Tech- nology and Systems17, 1–23 (2024) UNet−−:U-Net with Reduced Skip-Connections 17

    Weng, O., Marcano, G., Loncar, V., Khodamoradi, A., Sheybani, N., Meza, A., Koushanfar, F., Denolf, K., Duarte, J.M., Kastner, R.: Tailor: Altering skip connec- tions for resource-efficient inference. ACM Transactions on Reconfigurable Tech- nology and Systems17, 1–23 (2024) U...

    2024

  25. [33]

    arXiv preprint arXiv:2301.00808 (2023)

    Woo, S., Debnath, S., Hu, R., Chen, X., Liu, Z., Kweon, I.S., Xie, S.: Convnext v2: Co-designing and scaling convnets with masked autoencoders. arXiv preprint arXiv:2301.00808 (2023)

    2023 arXiv

  26. [34]

    Xie,S.,Girshick,R.,Dollár,P.,Tu,Z.,He,K.:Aggregatedresidualtransformations for deep neural networks (2017)

    2017

  27. [35]

    Multimedia Tools and Applications81, 15593–15607 (05 2022)

    Yang, X., Li, Z., Guo, Y., Zhou, D.: Dcu-net: a deformable convolutional neural network based on cascade u-net for retinal vessel segmentation. Multimedia Tools and Applications81, 15593–15607 (05 2022). https://doi.org/10.1007/s11042-022- 12418-w

    2022 doi

  28. [36]

    In: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Yu, Q., Zhang, J., Zhang, H., Wang, Y., Lin, Z., Xu, N., Bai, Y., Yuille, A.: Mask guided matting via progressive refinement network. In: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 1154–1163 (2021). https://doi.org/10.1109/CVPR46437.2021.00121

    2021

  29. [37]

    In: International Conference on Curves and Surfaces (2012)

    Zeyde, R., Elad, M., Protter, M.: On single image scale-up using sparse- representations. In: International Conference on Curves and Surfaces (2012)

    2012

  30. [38]

    In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion

    Zhang, H., Wu, C., Zhang, Z., Zhu, Y., Lin, H., Zhang, Z., Sun, Y., He, T., Mueller, J., Manmatha, R., Li, M., Smola, A.: Resnest: Split-attention networks. In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion. pp. 2736–2746 (2020)

    2020

  31. [39]

    Applied Intelligence52, 295–304 (Jan 2022)

    Zhang, Y., Sun, Y., Liu, S.: Deformable and residual convolutional network for image super-resolution. Applied Intelligence52, 295–304 (Jan 2022)

    2022

  32. [40]

    arXiv preprint arXiv:2303.02881 (2023)

    Zhang, Y., Li, D., Shi, X., He, D., Song, K., Wang, X., Qin, H., Li, H.: Kbnet: Kernel basis network for image restoration. arXiv preprint arXiv:2303.02881 (2023)

    2023 arXiv

  33. [41]

    Proceedings of the IEEE109(5), 820–838 (May 2021)

    Zhou, S.K., Greenspan, H., Davatzikos, C., Duncan, J.S., Van Ginneken, B., Mad- abhushi, A., Prince, J.L., Rueckert, D., Summers, R.M.: A review of deep learning in medical imaging: Imaging traits, technology trends, case studies with progress highlights, and future promises. ...

    2021

  34. [42]

    IEEE Transactions on Image Processing 13(4), 600–612 (2004)

    Zhou, W., Bovik, A., Sheikh, H., Simoncelli, E.: Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing 13(4), 600–612 (2004). https://doi.org/10.1109/TIP.2003.819861

    2004

  35. [43]

    Journal of IEEE Transactions on Medical Imaging39(2), 1856–1867 (2019)

    Zhou, Z., Siddiquee, M.M.R., Tajbakhsh, N., Liang, J.: Unet++: Redesigning skip connections to exploit multiscale features in image segmentation. Journal of IEEE Transactions on Medical Imaging39(2), 1856–1867 (2019)

    2019

  36. [44]

    Springer, Cham (2016)

    Özgün Çiçek, Abdulkadir, A., Lienkamp, S.S., Brox, T., Ronneberger, O.: 3d u-net: Learning dense volumetric segmentation from sparse annotation. Springer, Cham (2016)

    2016

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.