Pith. sign in

REVIEW 3 major objections 49 references

GASP: A Gradient-Aware Shortest Path Algorithm for Boundary-Confined Visualization of 2-Manifold Reeb Graphs

T0 review · 3 major / 0 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read GASP produces Reeb graph drawings that stay on the domain boundary, stay compact, and follow the scalar field's gradient, making them more faithful to the data than existing layouts.

desk verdict No GASP paper here: the full text is an unrelated action-segmentation submission, so the abstract alone cannot support a review; desk reject until the correct manuscript is uploaded. read the letter →

arxiv 2508.05524 v2 pith:WAQMNDT2 submitted 2025-08-07 cs.GR cs.CGcs.HC

Sefat E. Rahman , Tushar M. Athawale , Paul Rosen This is my paper · ORCID
classification cs.GRcs.CGcs.HC
keywords Reebgraphscalarfieldtopology2-manifoldgradientalignmentboundary-constrainedlayoutcompactdrawingToolKitvisualizationfaithfulness
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 GASP, a layout algorithm for Reeb graphs of scalar fields on 2-manifolds. It argues that a faithful Reeb graph drawing should satisfy three properties: vertices and edges constrained to the domain boundary, a compact footprint, and edges aligned with the scalar function's gradient. Existing drawing algorithms, such as the geometric barycenter approach in the Topology ToolKit (TTK), ignore or violate these properties. GASP is designed to meet all three at once, and the paper reports qualitative and quantitative comparisons showing its drawings are more representative of the underlying data. A reader should care because Reeb graphs are a standard topological summary of scalar fields, and their usefulness depends on the drawing conveying the field's geometry, not just its connectivity.

What carries the argument

The central object is the Reeb graph of a scalar function $f$ on a 2-manifold: the graph obtained by contracting each connected component of a level set to a point, so nodes mark topological changes in the level sets and edges trace how those components evolve. The machinery is GASP, a gradient-aware shortest-path algorithm, which places Reeb graph edges as paths that follow the direction of the scalar field's gradient while keeping the layout within the boundary and compact. This routing is what ties the three faithfulness properties together: edges do not merely connect nodes, they also encode the direction of change in the scalar field.

What would settle it

Construct a 2-manifold scalar field whose Reeb graph contains an edge connecting two boundary points whose gradient descent and ascent paths are separated by an interior ridge. On such an input, a boundary-confined, gradient-aligned route may not exist without lengthening the edge or introducing crossings; if GASP's output on that field either leaves the boundary, deviates from the gradient direction, or changes the graph structure, the central premise fails. A reader can test this by computing the average angular deviation between each drawn edge and the local gradient direction over a suite

Watch

Extended reading notes

Core claim

On its own terms, the paper's claim is that a Reeb graph visualization for a 2-manifold scalar field can and should be boundary-constrained, compact, and gradient-aligned, and that GASP achieves this by routing the graph with a gradient-aware shortest-path strategy. The authors identify these three properties as the criteria for a faithful representation, then demonstrate that GASP's output is qualitatively cleaner and quantitatively better than the geometric barycenter algorithm as implemented in TTK. If the claim holds, the spatial layout of a Reeb graph stops being an arbitrary bookkeeping device and becomes a readable statement about where the field's level sets collapse and how its valu

Load-bearing premise

The claim rests on the assumption that boundary confinement, compactness, and gradient alignment can always be satisfied together without distorting the Reeb graph's structure; if some fields force a trade-off among these properties, the faithfulness advantage of GASP narrows.

Editorial extensions

If this is right

  • A user of TTK can substitute GASP for the geometric barycenter draw and obtain Reeb graph layouts that stay inside the domain boundary and track the scalar function's gradient.
  • Reeb graph edges in GASP layouts carry directional meaning, so the drawing can serve as a visual proxy for gradient flow rather than a bare skeleton.
  • The three properties give a concrete, checkable standard for faithfulness that future Reeb graph layout algorithms can be measured against.
  • Because the evaluation is both qualitative and quantitative, adopting GASP is not just an aesthetic choice: the paper reports measurable agreement with the underlying data over the barycenter baseline.
  • Boundary confinement and compactness together reduce the chance that long edge detours or off-domain placement mislead a viewer about where level sets actually collapse.

Reading between the lines

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

  • If the three properties are treated as a definition of faithfulness, then GASP's drawings make edge direction a readable channel for gradient flow; the paper's wording stops at 'more representative,' but the natural reading is that the layout itself encodes the scalar field's geometry, not just its connectivity.
  • The same recipe should transfer to contour trees, merge trees, and higher-dimensional Reeb spaces, where boundary confinement is less natural but gradient alignment is still a meaningful layout criterion.
  • A testable extension is perceptual: compare whether analysts can infer the direction of increase of the field more accurately from GASP layouts than from barycenter layouts; the paper establishes the geometric improvement but leaves the human-readability gain implicit.
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 / 0 minor

Summary. The submission presents an abstract for "GASP: A Gradient-Aware Shortest Path Algorithm for Boundary-Confined Visualization of 2-Manifold Reeb Graphs," claiming an algorithm that produces Reeb graph drawings constrained to the boundary, compact, and aligned with the function gradient, with qualitative and quantitative comparison against the geometric barycenter implementation in the Topology ToolKit (TTK). However, the supplied full text is a completely unrelated paper, "Looking into the Unknown: Exploring Action Discovery for Segmentation of Known and Unknown Actions" (arXiv:2508.05529v1), which addresses temporal action segmentation in videos. The body contains no Reeb graph definitions, no GASP algorithm or pseudocode, no theoretical analysis, and no experiments on 2-manifold scalar fields or TTK. Consequently, none of the claims in the abstract are supported by the submitted manuscript.

Significance. If substantiated, a gradient-aware shortest-path method for Reeb graph layout that simultaneously achieves boundary confinement, compactness, and gradient alignment would be a useful contribution to topology-based visualization, particularly given TTK's widespread adoption. However, this submission provides no evidence for any of these claims. There is no algorithm specification, no complexity analysis, no formal statement of the three properties, and no evaluation protocol. The only strength that could be credited—such as reproducible code, machine-checked proofs, or parameter-free derivations—is entirely absent. As submitted, the manuscript cannot be assessed on its scientific merits because its body is a different paper.

major comments (3)
  1. [Full text (entire body)] The main text is the paper "Looking into the Unknown: Exploring Action Discovery for Segmentation of Known and Unknown Actions," not the Reeb graph visualization paper described in the abstract. It contains no mention of Reeb graphs, GASP, boundary confinement, compactness, gradient alignment, or the Topology ToolKit. The central claim that GASP produces more representative Reeb graph visualizations is therefore entirely unsupported by the submitted body.
  2. [Abstract, "evaluated both qualitatively and quantitatively"] The abstract promises a quantitative and qualitative comparison against TTK's geometric barycenter algorithm, but the body provides no evaluation protocol, metrics, datasets, or results for Reeb graph visualization. The experiments in the body concern action segmentation on Breakfast, 50Salads, and DesktopAssembly, with metrics such as MoF, Edit distance, and F1@{10,25,50}. There is no basis in the manuscript to assess superiority over the TTK baseline.
  3. [Abstract, "three properties for faithfully representing Reeb graphs"] The manuscript nowhere defines boundary confinement, compactness, or gradient alignment formally, nor does it analyze whether a layout satisfying all three properties is always realizable without distorting the graph's combinatorial structure. Since the body is unrelated, no definitions, theorems, or feasibility arguments are provided. This is a load-bearing gap: even the modeling premise of the abstract is not established in the submitted text.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found in the available text; GASP is evaluated against TTK's external geometric-barycenter baseline, and the mismatched full text leaves the derivation chain unwalkable but exhibits no specific circular reduction.

full rationale

The submission contains only the GASP abstract; the body is an unrelated Temporal Action Segmentation paper (Spurio et al.), so there is no GASP algorithm, equation, or metric definition to inspect. A circularity finding requires quoting the paper and exhibiting a specific reduction (Eq. X = Eq. Y by construction, a fitted parameter renamed as a prediction, or a load-bearing self-citation chain). No such reduction is available from the abstract alone. The abstract's three-property fidelity premise ('constrained to the boundary, compact, and aligned with the function gradient') is a definitional modeling choice about what faithful representation means; GASP being 'cognizant of these properties' and hence 'more representative' would be a tautology only if 'representative' were defined as exactly those properties and no external benchmark existed. But the abstract's stated evaluation is comparative: the layouts are 'evaluated both qualitatively and quantitatively against the geometric barycenter algorithm, using its implementation available in the Topology ToolKit (TTK)'. That external baseline provides an independent anchor — a showing that GASP beats the barycenter layout on the stated properties is not reducible to GASP's own objective. No parameter fitting, no prediction from fitted values, and no author self-citation is claimed in the abstract. The absence of the real full text is a completeness/integrity problem: the central algorithm and its evaluation cannot be verified, and correctness claims are unsupported by the submitted body. But an unsupported claim is a verification failure, not circularity; per the exhibition rule, no circular step can be identified from the text provided.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

Audit is impossible for arXiv:2508.05524 because the supplied full text is the body of a different paper (arXiv:2508.05529, Action Discovery for Temporal Action Segmentation). The only axioms identifiable are those stated or implied in the abstract. No free parameters or invented entities are visible at the abstract level.

assumptions (2)
  • domain assumption Faithfulness of a Reeb graph visualization is adequately captured by three properties: boundary confinement, compactness, and gradient alignment.
    Stated in the abstract: 'We have identified three properties for faithfully representing Reeb graphs in a visualization.' These are design criteria assumed to define fidelity, not derived. If other criteria matter, e.g., preserving relative edge lengths or planarity, the central improvement claim weakens.
  • standard math The upstream Reeb graph computation for a scalar function on a 2-manifold is correct and standard.
    The abstract concerns drawing and layout, not construction, of Reeb graphs; it presumes the usual critical-point-based construction of Reeb graphs for 2-manifolds. This is implied background rather than stated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GASP: A Gradient-Aware Shortest Path Algorithm for Boundary-Confined Visualization of 2-Manifold Reeb Graphs." pith.science (2026). https://pith.science/paper/WAQMNDT2

@misc{pith2026250805524,
  author       = {Pith},
  title        = {Pith review of: GASP: A Gradient-Aware Shortest Path Algorithm for Boundary-Confined Visualization of 2-Manifold Reeb Graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WAQMNDT2}},
  note         = {Machine review of arXiv:2508.05524}
}
read the original abstract

Reeb graphs are an important tool for abstracting and representing the topological structure of a function defined on a manifold. We have identified three properties for faithfully representing Reeb graphs in a visualization: they should be constrained to the boundary, compact, and aligned with the function gradient. Existing algorithms for drawing Reeb graphs are agnostic to or violate these properties. In this paper, we introduce an algorithm to generate Reeb graph visualizations, called GASP, that is cognizant of these properties, thereby producing visualizations that are more representative of the underlying data. To demonstrate the improvements, the resulting Reeb graphs are evaluated both qualitatively and quantitatively against the geometric barycenter algorithm, using its implementation available in the Topology ToolKit (TTK), a widely adopted tool for calculating and visualizing Reeb graphs.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 48 canonical work pages

  1. [1]

    MS-TCN: Multi- stage temporal convolutional network for action seg- mentation

    Yazan Abu Farha and Juergen Gall. MS-TCN: Multi- stage temporal convolutional network for action seg- mentation. In CVPR, 2019. 2, 3, 4, 5

    work page 2019

  2. [2]

    Unsupervised learning from narrated instruc- tion videos

    Jean-Baptiste Alayrac, Piotr Bojanowski, Nishant Agrawal, Josef Sivic, Ivan Laptev, and Simon Lacoste- Julien. Unsupervised learning from narrated instruc- tion videos. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2016. 5

    work page 2016

  3. [3]

    How much temporal long-term context is needed for action segmentation? In IEEE International Con- ference on Computer Vision (ICCV), 2023

    Emad Bahrami, Gianpiero Francesca, and Juergen Gall. How much temporal long-term context is needed for action segmentation? In IEEE International Con- ference on Computer Vision (ICCV), 2023. 1, 2

    work page 2023

  4. [4]

    Opental: Towards open set temporal action localization

    Wentao Bao, Qi Yu, and Yu Kong. Opental: Towards open set temporal action localization. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2979–2989, 2022. 2

    work page 2022

  5. [5]

    Exploiting vlm local- izability and semantics for open vocabulary action de- tection

    Wentao Bao, Kai Li, Yuxiao Chen, Deep Patel, Mar- tin Renqiang Min, and Yu Kong. Exploiting vlm local- izability and semantics for open vocabulary action de- tection. In 2025 IEEE/CVF Winter Conference on Ap- plications of Computer Vision (WACV) , pages 8291–

    work page 2025

  6. [6]

    Unified fully and timestamp supervised temporal action segmenta- tion via sequence to sequence translation

    Nadine Behrmann, S Alireza Golestaneh, Zico Kolter, J¨urgen Gall, and Mehdi Noroozi. Unified fully and timestamp supervised temporal action segmenta- tion via sequence to sequence translation. In Euro- pean Conference on Computer Vision , pages 52–68. Springer, 2022. 2, 4

    work page 2022

  7. [7]

    Quo vadis, ac- tion recognition? a new model and the kinetics dataset

    Joao Carreira and Andrew Zisserman. Quo vadis, ac- tion recognition? a new model and the kinetics dataset. In proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 6299–6308,

    work page

  8. [8]

    D3tw: Discriminative differ- entiable dynamic time warping for weakly supervised action alignment and segmentation

    Chien-Yi Chang, De-An Huang, Yanan Sui, Li Fei-Fei, and Juan Carlos Niebles. D3tw: Discriminative differ- entiable dynamic time warping for weakly supervised action alignment and segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3546–3555, 2019. 1, 2

    work page 2019

Show all 49 references
  1. [9]

    Leveraging action affinity and continuity for semi-supervised temporal action segmentation

    Guodong Ding and Angela Yao. Leveraging action affinity and continuity for semi-supervised temporal action segmentation. In European Conference on Com- puter Vision, pages 17–32. Springer, 2022. 2

    2022

  2. [10]

    Weakly-supervised ac- tion segmentation with iterative soft boundary assign- ment

    Li Ding and Chenliang Xu. Weakly-supervised ac- tion segmentation with iterative soft boundary assign- ment. In Proceedings of the IEEE Conference on Com- puter Vision and Pattern Recognition , pages 6508– 6516, 2018. 1, 2

    2018

  3. [11]

    Fast and unsupervised action boundary detec- tion for action segmentation

    Zexing Du, Xue Wang, Guoqing Zhou, and Qing Wang. Fast and unsupervised action boundary detec- tion for action segmentation. In Conference on Com- puter Vision and Pattern Recognition (CVPR) , pages 3313–3322, 2022. 3

    2022

  4. [12]

    Sct: Set constrained temporal transformer for set supervised action segmen- tation

    Mohsen Fayyaz and Jurgen Gall. Sct: Set constrained temporal transformer for set supervised action segmen- tation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 501– 510, 2020. 1, 2

    2020

  5. [13]

    Open-vocabulary temporal action localiza- tion using multimodal guidance

    Akshita Gupta, Aditya Arora, Sanath Narayan, Salman Khan, Fahad Shahbaz Khan, and Graham W Tay- lor. Open-vocabulary temporal action localiza- tion using multimodal guidance. arXiv preprint arXiv:2406.15556, 2024. 2

    2024 arXiv

  6. [14]

    Im- proving action segmentation via graph-based temporal reasoning

    Yifei Huang, Yusuke Sugano, and Yoichi Sato. Im- proving action segmentation via graph-based temporal reasoning. In CVPR, 2020. 2

    2020

  7. [15]

    Alleviating over-segmentation errors by detecting action boundaries

    Yuchi Ishikawa, Seito Kasai, Yoshimitsu Aoki, and Hi- rokatsu Kataoka. Alleviating over-segmentation errors by detecting action boundaries. In WACV, 2021. 2

    2021

  8. [16]

    Fast saliency based pooling of fisher encoded dense trajectories

    Svebor Karaman, Lorenzo Seidenari, and Alberto Del Bimbo. Fast saliency based pooling of fisher encoded dense trajectories. In ECCV Workshops, 2014. 2

    2014

  9. [17]

    Timestamp-supervised action segmentation with graph convolutional networks

    Hamza Khan, Sanjay Haresh, Awais Ahmed, Shakeeb Siddiqui, Andrey Konin, M Zeeshan Zia, and Quoc- Huy Tran. Timestamp-supervised action segmentation with graph convolutional networks. In 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 1061...

    2022

  10. [18]

    The lan- guage of actions: Recovering the syntax and semantics of goal-directed human activities

    Hilde Kuehne, Ali Arslan, and Thomas Serre. The lan- guage of actions: Recovering the syntax and semantics of goal-directed human activities. In CVPR, 2014. 4

    2014

  11. [19]

    An end-to-end generative framework for video segmenta- tion and recognition

    Hilde Kuehne, Juergen Gall, and Thomas Serre. An end-to-end generative framework for video segmenta- tion and recognition. In WACV, 2016. 2

    2016

  12. [20]

    Weakly supervised learning of actions from transcripts

    Hilde Kuehne, Alexander Richard, and Juergen Gall. Weakly supervised learning of actions from transcripts. Computer Vision and Image Understanding , 163:78– 89, 2017. 1, 2

    2017

  13. [21]

    Unsupervised learning of action classes with continuous temporal embedding

    Anna Kukleva, Hilde Kuehne, Fadime Sener, and Jur- gen Gall. Unsupervised learning of action classes with continuous temporal embedding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 12066–12074, 2019. 2, 5

    2019

  14. [22]

    Unsu- pervised action segmentation by joint representation learning and online clustering

    Sateesh Kumar, Sanjay Haresh, Awais Ahmed, Andrey Konin, M Zeeshan Zia, and Quoc-Huy Tran. Unsu- pervised action segmentation by joint representation learning and online clustering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 2...

    2022

  15. [23]

    Set-constrained viterbi for set-supervised action segmentation

    Jun Li and Sinisa Todorovic. Set-constrained viterbi for set-supervised action segmentation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10820–10829, 2020. 1, 2

    2020

  16. [24]

    Weakly super- vised energy-based learning for action segmentation

    Jun Li, Peng Lei, and Sinisa Todorovic. Weakly super- vised energy-based learning for action segmentation. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 6243–6251, 2019. 1, 2

    2019

  17. [25]

    MS-TCN++: Multi-stage temporal convolutional network for action segmenta- tion

    Shi-Jie Li, Yazan Abu Farha, Yun Liu, Ming-Ming Cheng, and Juergen Gall. MS-TCN++: Multi-stage temporal convolutional network for action segmenta- tion. TPAMI, 2020. 2

    2020

  18. [26]

    Tempo- ral action segmentation from timestamp supervision

    Zhe Li, Yazan Abu Farha, and Jurgen Gall. Tempo- ral action segmentation from timestamp supervision. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition , pages 8365– 8374, 2021. 1, 2

    2021

  19. [27]

    Diffusion ac- tion segmentation

    Daochang Liu, Qiyue Li, Anh-Dung Dinh, Tingting Jiang, Mubarak Shah, and Chang Xu. Diffusion ac- tion segmentation. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 10139–10149, 2023. 2

    2023

  20. [28]

    Weakly-supervised ac- tion segmentation and alignment via transcript-aware union-of-subspaces learning

    Zijia Lu and Ehsan Elhamifar. Weakly-supervised ac- tion segmentation and alignment via transcript-aware union-of-subspaces learning. In Proceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 8085–8095, 2021. 1, 2

    2021

  21. [29]

    Fact: Frame-action cross-attention temporal modeling for efficient action segmentation

    Zijia Lu and Ehsan Elhamifar. Fact: Frame-action cross-attention temporal modeling for efficient action segmentation. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pages 18175–18185, 2024. 1, 2

    2024

  22. [30]

    Streamer: Streaming representation learn- ing and event segmentation in a hierarchical manner

    Ramy Mounir, Sujal Vijayaraghavan, and Sudeep Sarkar. Streamer: Streaming representation learn- ing and event segmentation in a hierarchical manner. 8 In A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, editors, Advances in Neu- ral Information Processing S...

  23. [31]

    Representation learning with contrastive predictive coding

    Aaron van den Oord, Yazhe Li, and Oriol Vinyals. Representation learning with contrastive predictive coding. arXiv preprint arXiv:1807.03748, 2018. 3

    2018 arXiv

  24. [32]

    A generalized and robust framework for timestamp supervision in temporal action segmen- tation

    Rahul Rahaman, Dipika Singhania, Alexandre Thiery, and Angela Yao. A generalized and robust framework for timestamp supervision in temporal action segmen- tation. In Computer Vision–ECCV 2022: 17th Eu- ropean Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part...

    2022

  25. [33]

    Temporal action detection using a statistical language model

    Alexander Richard and Juergen Gall. Temporal action detection using a statistical language model. In CVPR,

  26. [34]

    Weakly supervised action learning with rnn based fine- to-coarse modeling

    Alexander Richard, Hilde Kuehne, and Juergen Gall. Weakly supervised action learning with rnn based fine- to-coarse modeling. In Proceedings of the IEEE Con- ference on Computer Vision and Pattern Recognition , pages 754–763, 2017. 1, 2

    2017

  27. [35]

    Action sets: Weakly supervised action segmentation without ordering constraints

    Alexander Richard, Hilde Kuehne, and Juergen Gall. Action sets: Weakly supervised action segmentation without ordering constraints. In Proceedings of the IEEE conference on Computer Vision and Pattern Recognition, pages 5987–5996, 2018. 1, 2

    2018

  28. [36]

    Neuralnetwork-viterbi: A framework for weakly supervised video learning

    Alexander Richard, Hilde Kuehne, Ahsan Iqbal, and Juergen Gall. Neuralnetwork-viterbi: A framework for weakly supervised video learning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 7386–7395, 2018. 1, 2

    2018

  29. [37]

    A database for fine grained activity detection of cooking activities

    Marcus Rohrbach, Sikandar Amin, Mykhaylo An- driluka, and Bernt Schiele. A database for fine grained activity detection of cooking activities. InCVPR, 2012. 2

    2012

  30. [38]

    Temporally- weighted hierarchical clustering for unsupervised ac- tion segmentation

    Saquib Sarfraz, Naila Murray, Vivek Sharma, Ali Diba, Luc Van Gool, and Rainer Stiefelhagen. Temporally- weighted hierarchical clustering for unsupervised ac- tion segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, pages 11225...

    2021

  31. [39]

    Estimating the dimension of a model

    Gideon Schwarz. Estimating the dimension of a model. Ann. Stat., 6(2):461–464, March 1978. 4

    1978

  32. [40]

    Unsupervised learning and segmentation of complex activities from video

    Fadime Sener and Angela Yao. Unsupervised learning and segmentation of complex activities from video. In Proceedings of the IEEE Conference on Computer Vi- sion and Pattern Recognition, pages 8368–8376, 2018. 5

    2018

  33. [41]

    Iterative contrast-classify for semi-supervised tempo- ral action segmentation

    Dipika Singhania, Rahul Rahaman, and Angela Yao. Iterative contrast-classify for semi-supervised tempo- ral action segmentation. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 2262–2270, 2022. 2

    2022

  34. [42]

    Robust action seg- mentation from timestamp supervision

    Yaser Souri, Yazan Abu Farha, Emad Bahrami, Gian- piero Francesca, and Juergen Gall. Robust action seg- mentation from timestamp supervision. arXiv preprint,

  35. [43]

    Hierarchical vector quantization for unsupervised action segmentation

    Federico Spurio, Emad Bahrami, Gianpiero Francesca, and Juergen Gall. Hierarchical vector quantization for unsupervised action segmentation. In Proceedings of the AAAI Conference on Artificial Intelligence , vol- ume 39, pages 6996–7005, 2025. 2

    2025

  36. [44]

    Sebastian Stein and Stephen J. McKenna. Combin- ing embedded accelerometers with computer vision for recognizing food preparation activities. InProceedings of the 2013 ACM International Joint Conference on Pervasive and Ubiquitous Computing , UbiComp ’13, page 729–738, New Yor...

    2013

  37. [45]

    Learning latent temporal structure for complex event detection

    Kevin Tang, Li Fei-Fei, and Daphne Koller. Learning latent temporal structure for complex event detection. In CVPR, 2012. 2

    2012

  38. [46]

    Permutation-aware activ- ity segmentation via unsupervised frame-to-segment alignment

    Quoc-Huy Tran, Ahmed Mehmood, Muhammad Ahmed, Muhammad Naufil, Anas Zafar, Andrey Konin, and Zeeshan Zia. Permutation-aware activ- ity segmentation via unsupervised frame-to-segment alignment. In Proceedings of the IEEE/CVF Win- ter Conference on Applications of Computer Visio...

    2024

  39. [47]

    Temporally consistent unbalanced optimal transport for unsupervised action segmentation

    Ming Xu and Stephen Gould. Temporally consistent unbalanced optimal transport for unsupervised action segmentation. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pages 14618–14627, 2024. 2, 6

    2024

  40. [48]

    AS- Former: Transformer for action segmentation

    Fangqiu Yi, Hongyu Wen, and Tingting Jiang. AS- Former: Transformer for action segmentation. In BMVC, 2021. 1, 2, 4

    2021

  41. [49]

    Ow- tal: learning unknown human activities for open-world temporal action localization

    Yaru Zhang, Xiao-Yu Zhang, and Haichao Shi. Ow- tal: learning unknown human activities for open-world temporal action localization. Pattern Recognition, 133: 109027, 2023. 2 9 Looking into the Unknown: Exploring Action Discovery for Segmentation of Known and Unknown Actions - ...

    2023

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.