Pith. sign in

REVIEW 4 major objections 5 minor 56 references

Back to Fundamentals: Low-Level Visual Features Guided Progressive Token Pruning

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A two-stage entropy-and-edge score prunes ViT tokens with no training, cutting compute by up to 46% while holding segmentation accuracy.

desk verdict A plausible new token-pruning recipe with transparent ablations, but the reference baseline shifts between tables and the SAM mIoU protocol is never specified, so the headline trade-off numbers are not yet reproducible. read the letter →

arxiv 2504.17996 v1 pith:NFDCV2FI submitted 2025-04-25 cs.CV

Yuanbing Ouyang , Yizhuo Liang , Qingpeng Li , Xinfei Guo , Yiming Luo , Di Wu , Hao Wang , Yushan Pan This is my paper · ORCID
classification cs.CV
keywords VisionTransformerstokenpruningTsallisentropysemanticsegmentationSobeledgedetectionplug-and-playcompressionmodelefficiencyedge-aware
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 introduces LVTP, a training-free token pruning method for Vision Transformers used in semantic segmentation. It claims that scoring tokens with multi-scale Tsallis entropy and then re-checking survivors with Sobel edge maps preserves segmentation accuracy far better than methods that rely on high-level semantics alone. On a frozen SAM (Segment Anything) ViT-H it reports a 46.2% GFLOPS reduction with a 5 point mIoU drop on RIO, and a 45.5% GFLOPS reduction with a 3 point mIoU drop on COCO-Tasks. If these numbers hold, LVTP offers a plug-and-play route to cheaper ViT-based segmentation on resource-constrained devices without retraining or architectural change.

What carries the argument

The load-bearing object is a two-scale Tsallis entropy score, a non-extensive generalization of Shannon entropy, written as $S_{q_1,q_2}(P)=\alpha\cdot(1-\sum_i p_i^{q_1})/(q_1-1)+\beta\cdot(1-\sum_i p_i^{q_2})/(q_2-1)$, with $q_1>1$ and $q_2<1$, and dynamic weights $\alpha=\|F\|_2/\sqrt{D}$, $\beta=1-\alpha$. This entropy weight matrix modulates token features before cosine-distance k-means clustering, and a second clustering is driven by the Sobel gradient magnitude $G(x,y)=G_x(x,y)^2+G_y(x,y)^2$. The machinery classifies tokens as redundant or informative at two complementary granularities: entropy catches semantic richness, while Sobel catches boundary and texture cues that high-level scoring would prune away.

What would settle it

Re-run the same no-training protocol on a class-labeled semantic segmentation benchmark such as ADE20K or Cityscapes with a ViT segmentor whose task head is fixed; if a 45% GFLOPS reduction comes with an mIoU drop larger than the claimed 3 points on COCO-style data, or if removing the Sobel-guided second clustering does not increase mIoU relative to a simple second clustering, the paper's central claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that token redundancy in ViT-based semantic segmentation can be removed in two stages. The first stage clusters tokens with an entropy-weighted feature representation using multi-scale Tsallis entropy, which scores semantic importance while balancing salient high-probability features ($q_1>1$) and detail-oriented low-probability features ($q_2<1$). The second stage refines that decision with Sobel edge maps so boundary tokens survive pruning. Across SAM ViT-H/ViT-L and Swin-Unet backbones on RIO, COCO-Tasks, and Massachusetts Road, this progressive scheme reduces GFLOPS by about 20% to 46% while keeping mIoU degradation between roughly 0 and 6 points, and it outperforms CTS, DToP, SViT, and VLTP on the gamma tradeoff metric, defined as the ratio of GFLOPS reduction to mIoU decrease, under a no-fine-tuning comparison.

Load-bearing premise

The central measurements assume SAM's outputs can be scored as class-level semantic segmentation mIoU on the RIO and COCO datasets without task adaptation, even though SAM is an interactive, class-agnostic segmenter; if its masks are not class-labeled semantic predictions, the reported mIoU drops do not measure semantic segmentation quality.

Editorial extensions

If this is right

  • LVTP is plug-and-play: it applies to frozen ViT, Swin, and SAM backbones with no architectural change and no fine-tuning.
  • Pruning at a middle transformer layer (layer 16 of SAM ViT-H) gives the best cost-accuracy tradeoff; pruning layer 8 destroys primitive features and pruning layer 24 disrupts global semantics.
  • The Sobel-guided second clustering is what preserves boundary accuracy: removing it raises the mIoU drop from 5.0 to 11.9 points on RIO.
  • Moderate pruning can even improve accuracy on RIO, where mIoU rises from 46.5% at 0% second-stage pruning to 48.5% at 40%.
  • LVTP has the best GFLOPS-to-mIoU tradeoff among the training-free token pruning methods compared in Table 1.

Reading between the lines

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

  • The Sobel-guided second stage only needs the input image, so the same two-stage scoring should transfer to other dense prediction tasks such as instance or panoptic segmentation; the paper lists this as future work, not a demonstrated result.
  • The RIO finding that accuracy improves as the second-stage pruning rate rises to 40% suggests entropy-guided pruning also acts as a mild regularizer by suppressing noisy background tokens, an effect the paper does not claim explicitly.
  • A testable extension is to replace the Sobel operator with learned boundary maps or structured edge detectors to see whether the extreme conditions named in the discussion, such as lighting changes, occlusion, and noise, can be handled without retraining the segmentor.
  • The gamma ratio combines a percentage GFLOPS change with a percentage-point mIoU change, so cross-method comparisons should be read alongside the absolute mIoU and GFLOPS numbers, not the ratio alone.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 LVTP, a training-free token pruning framework for ViT-based semantic segmentation. The method scores tokens with a multi-scale Tsallis entropy weighting, performs a first k-means clustering over entropy-weighted features, and then applies a second, Sobel-edge-guided clustering to preserve boundary-critical tokens before pruning at a selected transformer layer. The authors report GFLOPS reductions of roughly 20%-46% with small mIoU drops on RIO, COCO-Tasks, and Massachusetts Roads, and claim a superior trade-off (gamma) compared with CTS, DToP, SViT, and VLTP under a no-fine-tuning protocol. The paper also includes ablations over entropy choices, second-clustering variants, pruning rates, and pruning layers.

Significance. If the central claims were reliable, the paper would offer a practically attractive plug-in: a no-training, architecture-agnostic token pruning method that uses low-level image structure to retain segmentation boundaries. The motivation is reasonable, and the component-level ablations (Tables 3, 4, and 6) are a useful step toward understanding what drives the behavior. However, the current validation is not trustworthy because the baseline configuration is inconsistent across tables, the use of SAM for semantic segmentation mIoU is not specified, and the headline gamma is selected on the same benchmarks used for evaluation. These issues affect the paper's main quantitative claims rather than only its presentation, so the manuscript is not yet suitable for publication.

major comments (4)
  1. [Table 2 and Table 5] The reference configuration for the central claim is inconsistent. For SAM ViT-H on RIO, Table 2 lists a baseline of 820 GFLOPS and 0.53 mIoU, while Table 5 lists a '0% pruning rate' row with 528 GFLOPS and 46.5% mIoU. The 50% rows agree (441 GFLOPS; 0.48/48.2 mIoU), but the reference points do not. The paper does not state whether the 0% row is the untouched model or the model after first-stage entropy clustering. If it is the latter, calling it '0% pruning rate' is misleading; if it is the former, the same configuration is measured with a 35% GFLOPS difference. Either way, the reported -46.2% GFLOPS and gamma = 9.24 in Table 1 are not anchored to a well-specified baseline.
  2. [Section 4.1 and Table 2] SAM is a class-agnostic, promptable segmentation model, yet the paper reports semantic segmentation mIoU on COCO-Tasks and RIO without describing how class-level semantic masks are derived. There is no discussion of prompts, class mapping, or any task adaptation. Unless this protocol is specified, the mIoU numbers do not measure semantic segmentation as understood in the comparison methods (CTS, DToP, SViT), and the comparison in Table 1 is not valid.
  3. [Eq. (11) and Tables 1, 2, 5] The metric gamma is defined as the ratio of the reduction ratio in GFLOPS to the percentage decrease in mIoU, but the units of mIoU are used inconsistently. Table 2 reports mIoU as proportions (0.53, 0.48), while Table 5 reports percentages (46.5%, 48.2%). The claimed -5% mIoU change from 0.53 to 0.48 is an absolute 5-percentage-point drop, not a relative 9.4% decrease. The gamma values therefore depend on an unstated and non-uniform interpretation of Delta mIoU, making them incomparable across tables.
  4. [Tables 5 and 6 vs. Table 1] The headline gamma is computed after selecting the pruning layer (16th layer) and the second-clustering pruning rate (50%) from a scan over the same benchmarks in Tables 5 and 6. This is a selection-on-the-test-set procedure, so the reported superiority of LVTP in Table 1 partly reflects the choice of the best configuration rather than an independent evaluation. A held-out validation split or a fixed default configuration should be used to report the main result.
minor comments (5)
  1. [Section 3.2-3.3] The values of q1 and q2, the number of clusters k, and the initialization of cluster centers are not reported, even though these are central free parameters of the method; a sensitivity study or at least the chosen values is needed for reproducibility.
  2. [Figure 5 caption] Figure 5 is captioned 'Framework for Edge-Enhanced Token Clustering in Transformers', which is the same title as Figure 1; the caption should describe the gamma-versus-pruning-rate visualization instead.
  3. [Section 4.4.2, Table 4] In the text before Table 4, the RIO Sobel-guided result is described as a GFLOPS reduction of '-48.2%', but the table lists '-46.2%'; the numbers should be reconciled.
  4. [Section 4.4.3, Table 5] The paper should state explicitly whether the '0%' row in Table 5 already includes the first-stage entropy-based clustering; without this clarification the table cannot be interpreted as a pruning-rate ablation.
  5. [Throughout] There are numerous typographical and language errors ('focued', 'becames', 'di fferent', inconsistent article usage) that should be corrected in a revision.

Circularity Check

1 steps flagged · score 4.0 of 10

Headline SOTA gamma is a post-selection measurement: the pruning rate and layer are chosen from a scan over the same benchmarks, so part of the claimed superiority is produced by the selection procedure rather than independently predicted.

  1. fitted input called prediction [Section 4.3.1 (Table 1), Section 4.4.3 (Table 5), Section 4.4.4 (Table 6)]
    "Combining the above experiment results, the advantage of a 50% pruning rate was evident ... a 50% pruning rate could achieve the best balance between computational reduction and performance. As shown in Table 6, pruning the 16th layer achieved the best balance between reducing computational cost and maintaining model performance, delivering the optimal γ on both the COCO and RIO datasets (9.24 and 14.7, respectively)."

    The headline γ=9.24 is not an independent prediction. The 50% pruning rate and 16th layer are selected from an ablation scan over the same RIO/COCO benchmarks using the same γ criterion, and the reported value is the maximum of that scan. Thus the claim that LVTP 'significantly surpasses' competitors is a restatement of the selection rule: the configuration is chosen to maximize the very metric that is then cited as evidence, so the central SOTA comparison is statistically forced by the selection procedure rather than derived from the entropy/Sobel mechanism.

full rationale

LVTP's core mechanism is largely self-contained: Equations (3)-(10) define a training-free heuristic (multi-scale Tsallis weighting plus Sobel-guided second clustering), and the reported GFLOPS/mIoU reductions are empirical measurements, not predictions derived from the equations. The ablations in Tables 3 and 4 provide independent evidence that the entropy and Sobel components contribute. There is no load-bearing self-citation: the cited prior work is contextual, and no uniqueness theorem is imported. However, the central SOTA claim is weakened by a post-selection effect: the optimal pruning rate and layer are chosen from the same benchmarks on which the headline γ is reported, so the advantage is partly an artifact of the selection procedure. Separately, the reference configuration is unstable—Table 2 reports 820 GFLOPS/0.53 mIoU for SAM-RIO ViT-H while Table 5's 0% row reports 528 GFLOPS/46.5% mIoU—which is a reproducibility concern rather than a definitional circularity, but it further undermines the anchoring of ΔGFLOPS/ΔmIoU. Overall, partial circularity in the headline comparison, with independent content in the mechanism ablations.

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

The central method depends on several tuned or unreported parameters (q1, q2, pruning rate, pruning layer, cluster count) and on unverified modeling assumptions about entropy, clustering, Sobel edges, and the suitability of SAM and Swin-Unet for the chosen benchmarks. No new physical entities are introduced.

free parameters (5)
  • q1 (Tsallis saliency exponent) = unspecified (stated q1 > 1)
    Controls high-probability feature emphasis in Eq. (3); value is never reported or swept.
  • q2 (Tsallis detail exponent) = unspecified (stated q2 < 1)
    Controls low-probability detail emphasis; value is not reported.
  • Second-clustering pruning rate = 50% in main tables
    Selected from ablation Table 5; the method does not determine it, and on RIO a 40% rate gives higher mIoU (48.5) than 50% (48.2).
  • Pruning layer = 16th (ViT-H), 12th (ViT-L), 2nd (Swin-L/B)
    Selected from the Table 6 ablation; not intrinsic to the method.
  • k in k-means clustering = unspecified
    The cluster count for both clustering rounds is never given, though clustering is central to the method.
assumptions (5)
  • domain assumption Token redundancy: over 60% of tokens in natural images contribute little to prediction.
    Motivates pruning; cited to prior work and not verified for the specific datasets used here.
  • ad hoc to paper Tsallis entropy with q1 > 1 and q2 < 1 captures token semantic importance.
    Stated in Section 3.2 without derivation or external justification.
  • ad hoc to paper Entropy-weighted k-means with cosine similarity yields clusters that separate semantically important and redundant tokens.
    Section 3.3, no analysis of clustering quality or convergence.
  • ad hoc to paper Sobel edge magnitude identifies tokens whose preservation is necessary for segmentation accuracy.
    Section 3.4, a heuristic with no formal justification linking edge magnitude to mIoU preservation.
  • domain assumption SAM and Swin-Unet can be applied directly as semantic segmentation models on RIO/COCO/Massachusetts without task-specific adaptation.
    Underlies all mIoU numbers; the paper does not describe how class-level masks are produced from these models.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Back to Fundamentals: Low-Level Visual Features Guided Progressive Token Pruning." pith.science (2026). https://pith.science/paper/NFDCV2FI

@misc{pith2026250417996,
  author       = {Pith},
  title        = {Pith review of: Back to Fundamentals: Low-Level Visual Features Guided Progressive Token Pruning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NFDCV2FI}},
  note         = {Machine review of arXiv:2504.17996}
}
read the original abstract

Vision Transformers (ViTs) excel in semantic segmentation but demand significant computation, posing challenges for deployment on resource-constrained devices. Existing token pruning methods often overlook fundamental visual data characteristics. This study introduces 'LVTP', a progressive token pruning framework guided by multi-scale Tsallis entropy and low-level visual features with twice clustering. It integrates high-level semantics and basic visual attributes for precise segmentation. A novel dynamic scoring mechanism using multi-scale Tsallis entropy weighting overcomes limitations of traditional single-parameter entropy. The framework also incorporates low-level feature analysis to preserve critical edge information while optimizing computational cost. As a plug-and-play module, it requires no architectural changes or additional training. Evaluations across multiple datasets show 20%-45% computational reductions with negligible performance loss, outperforming existing methods in balancing cost and accuracy, especially in complex edge regions.

Figures

Figures reproduced from arXiv: 2504.17996 by the authors.

Figure 1
Figure 1. Framework for Edge-Enhanced Token Clustering in Transformers [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Edge Enhancement Example Figure note: Image (a) shows the original grayscale image. Image (b) is the result of applying the Sobel operator to the original image, highlighting edges and making them more prominent. The red box in image (b) indicates a region of interest where edge enhancement is particularly noticeable. Image (c) is a zoomed-in view of the region within the red box, showing the detailed edge structure… view at source ↗
Figure 3
Figure 3. Pruned SAM-ViT-H’s prediction from RIO and COCO dataset [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Pruned Swin-Unet-L’s prediction from Massachusetts-Road dataset [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: Framework for Edge-Enhanced Token Clustering in Transformers [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 36 canonical work pages

  1. [55]

    K-means-type algorithms: A generalized convergence theorem and char- acterization of local optimality

    Shokri Z Selim and Mohamed A Ismail. K-means-type algorithms: A generalized convergence theorem and char- acterization of local optimality. IEEE Transactions on pattern analysis and machine intelligence, (1):81–87, 1984

    work page 1984

  2. [1]

    Improving language understanding by generative pre-training

    Alec Radford, Karthik Narasimhan, Tim Salimans, Ilya Sutskever, et al. Improving language understanding by generative pre-training. 2018

    2018

  3. [2]

    Language models are unsupervised multitask learners

    Alec Radford, Je ffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. Language models are unsupervised multitask learners. OpenAI blog, 1(8):9, 2019

    2019

  4. [3]

    Language models are few-shot learners

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Nee- lakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901, 2020

    1901

  5. [4]

    Transformers in vision: A survey

    Salman Khan, Muzammal Naseer, Munawar Hayat, Syed Waqas Zamir, Fahad Shahbaz Khan, and Mubarak Shah. Transformers in vision: A survey. ACM computing surveys (CSUR), 54(10s):1–41, 2022

    2022

  6. [7]

    A survey of visual transformers

    Yang Liu, Yao Zhang, Yixin Wang, Feng Hou, Jin Yuan, Jiang Tian, Yang Zhang, Zhongchao Shi, Jianping Fan, and Zhiqiang He. A survey of visual transformers. IEEE Transactions on Neural Networks and Learning Systems, 2023

    work page 2023

  7. [8]

    Stand-alone self-attention in vision models

    Prajit Ramachandran, Niki Parmar, Ashish Vaswani, Irwan Bello, Anselm Levskaya, and Jon Shlens. Stand-alone self-attention in vision models. Advances in neural information processing systems, 32, 2019

    work page 2019

  8. [9]

    Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs

    Liang-Chieh Chen, George Papandreou, Iasonas Kokkinos, Kevin Murphy, and Alan L Yuille. Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs. IEEE transactions on pattern analysis and machine intelligence, 40(4):834–848, 2017

    work page 2017

Show all 56 references
  1. [10]

    Transunet: Transformers make strong encoders for medical image segmentation

    Jieneng Chen, Yongyi Lu, Qihang Yu, Xiangde Luo, Ehsan Adeli, Yan Wang, Le Lu, Alan L Yuille, and Yuyin Zhou. Transunet: Transformers make strong encoders for medical image segmentation. arXiv preprint arXiv:2102.04306, 2021

    2021 arXiv

  2. [12]

    A survey on efficient vision transformers: algorithms, techniques, and performance benchmarking

    Lorenzo Papa, Paolo Russo, Irene Amerini, and Luping Zhou. A survey on efficient vision transformers: algorithms, techniques, and performance benchmarking. IEEE Transactions on Pattern Analysis and Machine Intelligence , 2024

    2024

  3. [13]

    Neural window fully-connected crfs for monocular depth estimation

    Weihao Yuan, Xiaodong Gu, Zuozhuo Dai, Siyu Zhu, and Ping Tan. Neural window fully-connected crfs for monocular depth estimation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recogni- tion, pages 3916–3925, 2022

    2022

  4. [14]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv...

    2010 arXiv

  5. [16]

    A survey on vision transformer

    Kai Han, Yunhe Wang, Hanting Chen, Xinghao Chen, Jianyuan Guo, Zhenhua Liu, Yehui Tang, An Xiao, Chun- jing Xu, Yixing Xu, et al. A survey on vision transformer. IEEE transactions on pattern analysis and machine intelligence, 45(1):87–110, 2022

    2022

  6. [17]

    Lstm: A search space odyssey

    Klaus Gre ff, Rupesh K Srivastava, Jan Koutn ´ık, Bas R Steunebrink, and J ¨urgen Schmidhuber. Lstm: A search space odyssey. IEEE transactions on neural networks and learning systems, 28(10):2222–2232, 2016. 17

    2016

  7. [18]

    Learned token pruning for transformers

    Sehoon Kim, Sheng Shen, David Thorsley, Amir Gholami, Woosuk Kwon, Joseph Hassoun, and Kurt Keutzer. Learned token pruning for transformers. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 784–794, 2022

    2022

  8. [19]

    Baseg: Boundary aware semantic segmentation for autonomous driving

    Xiaoyang Xiao, Yuqian Zhao, Fan Zhang, Biao Luo, Lingli Yu, Baifan Chen, and Chunhua Yang. Baseg: Boundary aware semantic segmentation for autonomous driving. Neural Networks, 157:460–470, 2023

    2023

  9. [20]

    Sait: Sparse vision transformers through adaptive token pruning

    Ling Li, David Thorsley, and Joseph Hassoun. Sait: Sparse vision transformers through adaptive token pruning. arXiv preprint arXiv:2210.05832, 2022

    2022 arXiv

  10. [21]

    Cross-image pixel contrasting for semantic segmentation

    Tianfei Zhou and Wenguan Wang. Cross-image pixel contrasting for semantic segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  11. [22]

    Segformer: Simple and efficient design for semantic segmentation with transformers

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M Alvarez, and Ping Luo. Segformer: Simple and efficient design for semantic segmentation with transformers. Advances in neural information processing systems, 34:12077–12090, 2021

    2021

  12. [24]

    Vltp: Vision-language guided token pruning for task-oriented segmentation

    Hanning Chen, Yang Ni, Wenjun Huang, Yezi Liu, SungHeon Jeong, Fei Wen, Nathaniel Bastian, Hugo Latapie, and Mohsen Imani. Vltp: Vision-language guided token pruning for task-oriented segmentation. arXiv preprint arXiv:2409.08464, 2024

    2024 arXiv

  13. [25]

    A survey on instance segmentation: state of the art.International journal of multimedia information retrieval, 9(3):171–189, 2020

    Abdul Mueed Hafiz and Ghulam Mohiuddin Bhat. A survey on instance segmentation: state of the art.International journal of multimedia information retrieval, 9(3):171–189, 2020

    2020

  14. [26]

    Yolact: Real-time instance segmentation

    Daniel Bolya, Chong Zhou, Fanyi Xiao, and Yong Jae Lee. Yolact: Real-time instance segmentation. In Proceed- ings of the IEEE/CVF international conference on computer vision, pages 9157–9166, 2019

    2019

  15. [27]

    Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images

    Ali Hatamizadeh, Vishwesh Nath, Yucheng Tang, Dong Yang, Holger R Roth, and Daguang Xu. Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images. In International MICCAI brainlesion workshop, pages 272–284. Springer, 2021

    2021

  16. [29]

    Decoupling foreground and background with siamese vit networks for weakly-supervised semantic segmentation

    Meiling Lin, Gongyan Li, Shaoyun Xu, Yuexing Hao, and Shu Zhang. Decoupling foreground and background with siamese vit networks for weakly-supervised semantic segmentation. Neurocomputing, 610:128540, 2024

    2024

  17. [31]

    Segvit v2: Exploring efficient and continual semantic segmentation with plain vision transformers

    Bowen Zhang, Liyang Liu, Minh Hieu Phan, Zhi Tian, Chunhua Shen, and Yifan Liu. Segvit v2: Exploring efficient and continual semantic segmentation with plain vision transformers. International Journal of Computer Vision, 132(4):1126–1147, 2024

    2024

  18. [32]

    Sam 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Ro- man R¨adle, Chloe Rolland, Laura Gustafson, et al. Sam 2: Segment anything in images and videos. arXiv preprint arXiv:2408.00714, 2024

    2024 arXiv

  19. [33]

    Semantic layering in room segmentation via llms

    Taehyeon Kim and Byung-Cheol Min. Semantic layering in room segmentation via llms. In 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 9831–9838. IEEE, 2024

    2024

  20. [34]

    Post-training quantization or quantization-aware training? that is the question

    Xiaotian Zhao, Ruge Xu, and Xinfei Guo. Post-training quantization or quantization-aware training? that is the question. In 2023 China Semiconductor Technology International Conference (CSTIC), pages 1–3. IEEE, 2023

    2023

  21. [35]

    Q-bert: Hessian based ultra low precision quantization of bert

    Sheng Shen, Zhen Dong, Jiayu Ye, Linjian Ma, Zhewei Yao, Amir Gholami, Michael W Mahoney, and Kurt Keutzer. Q-bert: Hessian based ultra low precision quantization of bert. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 8815–8821, 2020

    2020

  22. [36]

    Hawq: Hessian aware quan- tization of neural networks with mixed-precision

    Zhen Dong, Zhewei Yao, Amir Gholami, Michael W Mahoney, and Kurt Keutzer. Hawq: Hessian aware quan- tization of neural networks with mixed-precision. In Proceedings of the IEEE /CVF international conference on computer vision, pages 293–302, 2019

    2019

  23. [37]

    Post-training quantization for vision transformer

    Zhenhua Liu, Yunhe Wang, Kai Han, Wei Zhang, Siwei Ma, and Wen Gao. Post-training quantization for vision transformer. Advances in Neural Information Processing Systems, 34:28092–28103, 2021

    2021

  24. [39]

    Towards accurate post-training quantization for vision transformer

    Yifu Ding, Haotong Qin, Qinghua Yan, Zhenhua Chai, Junjie Liu, Xiaolin Wei, and Xianglong Liu. Towards accurate post-training quantization for vision transformer. InProceedings of the 30th ACM international conference on multimedia, pages 5380–5388, 2022

    2022

  25. [40]

    A simple and effective pruning approach for large language models

    Mingjie Sun, Zhuang Liu, Anna Bair, and J Zico Kolter. A simple and effective pruning approach for large language models. arXiv preprint arXiv:2306.11695, 2023

    2023 arXiv

  26. [41]

    Prune and tune: Improving e fficient pruning 18 techniques for massive language models

    Aaquib Syed, Phillip Huang Guo, and Vijaykaarti Sundarapandiyan. Prune and tune: Improving e fficient pruning 18 techniques for massive language models. 2023

    2023

  27. [42]

    Are sixteen heads really better than one? Advances in neural information processing systems, 32, 2019

    Paul Michel, Omer Levy, and Graham Neubig. Are sixteen heads really better than one? Advances in neural information processing systems, 32, 2019

    2019

  28. [43]

    Reducing transformer depth on demand with structured dropout

    Angela Fan, Edouard Grave, and Armand Joulin. Reducing transformer depth on demand with structured dropout. arXiv preprint arXiv:1909.11556, 2019

    1909 arXiv

  29. [44]

    Fq-vit: Post-training quantization for fully quantized vision transformer

    Yang Lin, Tianyu Zhang, Peiqin Sun, Zheng Li, and Shuchang Zhou. Fq-vit: Post-training quantization for fully quantized vision transformer. arXiv preprint arXiv:2111.13824, 2021

    2021 arXiv

  30. [45]

    Comprehensive survey of model compres- sion and speed up for vision transformers

    Feiyang Chen, Ziqian Luo, Lisang Zhou, Xueting Pan, and Ying Jiang. Comprehensive survey of model compres- sion and speed up for vision transformers. arXiv preprint arXiv:2404.10407, 2024

    2024 arXiv

  31. [46]

    Not all patches are what you need: Expediting vision transformers via token reorganizations

    Youwei Liang, Chongjian Ge, Zhan Tong, Yibing Song, Jue Wang, and Pengtao Xie. Not all patches are what you need: Expediting vision transformers via token reorganizations. arXiv preprint arXiv:2202.07800, 2022

    2022 arXiv

  32. [47]

    Adaptive token sampling for e fficient vision trans- formers

    Mohsen Fayyaz, Soroush Abbasi Koohpayegani, Farnoush Rezaei Jafari, Sunando Sengupta, Hamid Reza Vaezi Joze, Eric Sommerlade, Hamed Pirsiavash, and J ¨urgen Gall. Adaptive token sampling for e fficient vision trans- formers. In European Conference on Computer Vision, pages 396...

    2022

  33. [48]

    Spvit: Enabling faster vision transformers via latency-aware soft token pruning

    Zhenglun Kong, Peiyan Dong, Xiaolong Ma, Xin Meng, Wei Niu, Mengshu Sun, Xuan Shen, Geng Yuan, Bin Ren, Hao Tang, et al. Spvit: Enabling faster vision transformers via latency-aware soft token pruning. InEuropean conference on computer vision, pages 620–640. Springer, 2022

    2022

  34. [50]

    Ia-red2: Interpretability-aware redundancy reduction for vision transformers

    Bowen Pan, Rameswar Panda, Yifan Jiang, Zhangyang Wang, Rogerio Feris, and Aude Oliva. Ia-red2: Interpretability-aware redundancy reduction for vision transformers. Advances in neural information processing systems, 34:24898–24911, 2021

    2021

  35. [51]

    Token merging: Your vit but faster

    Daniel Bolya, Cheng-Yang Fu, Xiaoliang Dai, Peizhao Zhang, Christoph Feichtenhofer, and Judy Hoffman. Token merging: Your vit but faster. arXiv preprint arXiv:2210.09461, 2022

    2022 arXiv

  36. [52]

    Tokenlearner: What can 8 learned tokens do for images and videos? arXiv preprint arXiv:2106.11297, 2021

    Michael S Ryoo, AJ Piergiovanni, Anurag Arnab, Mostafa Dehghani, and Anelia Angelova. Tokenlearner: What can 8 learned tokens do for images and videos? arXiv preprint arXiv:2106.11297, 2021

    2021 arXiv

  37. [53]

    A-vit: Adaptive tokens for efficient vision transformer

    Hongxu Yin, Arash Vahdat, Jose M Alvarez, Arun Mallya, Jan Kautz, and Pavlo Molchanov. A-vit: Adaptive tokens for efficient vision transformer. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10809–10818, 2022

    2022

  38. [54]

    Dynamicvit: E fficient vision transformers with dynamic token sparsification

    Yongming Rao, Wenliang Zhao, Benlin Liu, Jiwen Lu, Jie Zhou, and Cho-Jui Hsieh. Dynamicvit: E fficient vision transformers with dynamic token sparsification. Advances in neural information processing systems , 34:13937– 13949, 2021

    2021

  39. [56]

    Multilevel image thresholding based on 2d histogram and maximum tsallis entropy—a differential evolution approach

    Soham Sarkar and Swagatam Das. Multilevel image thresholding based on 2d histogram and maximum tsallis entropy—a differential evolution approach. IEEE transactions on Image Processing, 22(12):4788–4797, 2013

    2013

  40. [57]

    Edge detection using guided sobel image filtering

    Rakesh Ranjan and Vinay Avasthi. Edge detection using guided sobel image filtering. Wireless Personal Commu- nications, 132(1):651–677, 2023

    2023

  41. [58]

    Machine learning for aerial image labeling

    V olodymyr Mnih. Machine learning for aerial image labeling. University of Toronto (Canada), 2013

    2013

  42. [59]

    What object should i use?-task driven object detection

    Johann Sawatzky, Yaser Souri, Christian Grund, and Jurgen Gall. What object should i use?-task driven object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7605– 7614, 2019

    2019

  43. [60]

    Rio: A benchmark for reasoning intention-oriented objects in open environments

    Mengxue Qu, Yu Wu, Wu Liu, Xiaodan Liang, Jingkuan Song, Yao Zhao, and Yunchao Wei. Rio: A benchmark for reasoning intention-oriented objects in open environments. Advances in Neural Information Processing Systems , 36:43041–43056, 2023

    2023

  44. [61]

    Swin-unet: Unet-like pure transformer for medical image segmentation

    Hu Cao, Yueyue Wang, Joy Chen, Dongsheng Jiang, Xiaopeng Zhang, Qi Tian, and Manning Wang. Swin-unet: Unet-like pure transformer for medical image segmentation. In European conference on computer vision , pages 205–218. Springer, 2022

    2022

  45. [62]

    Segment anything

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. Segment anything. In Proceedings of the IEEE/CVF interna- tional conference on computer vision, pages 4015–4026, 2023

    2023

  46. [63]

    Content-aware token sharing for e fficient semantic segmen- tation with vision transformers

    Chenyang Lu, Daan de Geus, and Gijs Dubbelman. Content-aware token sharing for e fficient semantic segmen- tation with vision transformers. In Proceedings of the IEEE /CVF conference on computer vision and pattern recognition, pages 23631–23640, 2023

    2023

  47. [64]

    Dynamic token pruning in plain vision transform- ers for semantic segmentation

    Quan Tang, Bowen Zhang, Jiajun Liu, Fagui Liu, and Yifan Liu. Dynamic token pruning in plain vision transform- ers for semantic segmentation. In Proceedings of the IEEE /CVF International Conference on Computer Vision , pages 777–786, 2023

    2023

  48. [65]

    Revisiting token pruning 19 for object detection and instance segmentation

    Yifei Liu, Mathias Gehrig, Nico Messikommer, Marco Cannici, and Davide Scaramuzza. Revisiting token pruning 19 for object detection and instance segmentation. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 2658–2668, 2024. 20

    2024

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.