Pith. sign in

REVIEW 4 major objections 6 minor 46 references

FSBench: A Figure Skating Benchmark for Advancing Artistic Sports Understanding

T0 review · 4 major / 6 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 The paper claims that current large language models understand figure skating poorly, and that instruction-tuning on FSAnno's fine-grained, multi-modal annotations substantially improves their performance, with the resulting model…

desk verdict A genuinely useful multi-modal figure skating benchmark, but the reported instruction-tuning gain is not convincingly separated from the GPT-4-generated training and evaluation pipeline. read the letter →

arxiv 2504.19514 v1 pith:DOM3GBWU submitted 2025-04-28 cs.CV

Rong Gao , Xin Liu , Zhuozhao Hu , Bohao Xing , Baiqiang Xia , Zitong Yu , Heikki Kälviäinen This is my paper · ORCID
classification cs.CV
keywords figureskatingartisticsportsbenchmarkmultimodalLLMinstructiontuningmotionunderstandingactionqualityassessmentdataset
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 FSAnno, a large-scale figure skating dataset with fine-grained annotations at three levels—prior knowledge, individual actions, and whole performances—and FSBench, an evaluation benchmark built from it. The authors' central claim is that current large language models understand artistic sports such as figure skating poorly, and that instruction-tuning on FSAnno's fine-grained, multi-modal annotations substantially improves their ability to describe and analyze skating elements. They support this claim with initial tests showing low accuracy on figure-skating rules and event questions, and with an instruction-tuned model whose captioning quality, measured by the AutoDQ metric, rises sharply compared with its base motion-language model. The intended contribution is a reusable resource: a dataset and benchmark that treat technical execution and artistic expression as connected rather than separate tasks.

What carries the argument

The central object is FSAnno combined with FSBench: a dataset and benchmark built from 783 complete performances across eleven Grand Prix and Junior Grand Prix events, spanning four program types and including negative samples from younger and less experienced skaters. The load-bearing mechanism is the multi-source annotation pipeline that fuses official judging reports (element categories, GOE scores, TES/PCS), transcribed commentator speech aligned to action timestamps, and LLM-synthesized summaries into a single multi-level, multi-modal resource. The motion and skeleton modalities, extracted from raw video and stripped of identifying appearance, are what allow the benchmark to evaluate understanding of movement quality while protecting athlete privacy and reducing reliance on prior knowledge of specific competitions. Instruction-tuning a motion-language model on this resource yields SkateLLM, the model whose improved captions demonstrate the dataset's value.

What would settle it

Have a panel of human figure-skating officials re-annotate a random sample of FSBench elements: the element category, GOE score, segmentation boundaries, and the alignment of commentary to the correct action. If agreement between the human labels and FSAnno's automatically generated labels is no better than chance or is systematically biased, then model scores on FSBench and SkateLLM's gains could reflect the annotation pipeline rather than genuine understanding of figure skating.

Watch

Extended reading notes

Core claim

FSBench is presented as the first benchmark for figure skating that spans technical and artistic understanding at multiple granularities, with FSAnno providing the underlying annotations. The benchmark contains FSBench-Text, roughly 4,200 multiple-choice questions on rules and event facts, and FSBench-Motion, which pairs motion and skeleton data with question-answer sets for six tasks: action recognition, single-action assessment, single-action commentary, action segmentation, whole-performance scoring across seven dimensions, and whole-performance commentary. The paper reports that existing LLMs score far below expert level on the knowledge questions, and that motion-based models without figure-skating instruction tuning produce poor captions. The paper's constructive claim is that fine-grained, multi-modal annotations—grounded in official judge reports, transcribed commentator audio, and LLM-synthesized performance summaries—are sufficient to teach a motion-language model noticeably more professional and technically grounded skating descriptions.

Load-bearing premise

The benchmark's scores and the reported improvement of SkateLLM depend on the assumption that the automatically produced labels—extracted movement data, transcribed commentary, and machine-written summaries—agree with what human judges and commentators actually said and scored.

Editorial extensions

If this is right

  • FSBench provides a six-task evaluation protocol that other researchers can use to compare models on figure skating, with a public training/test split and an official benchmark split.
  • The benchmark's results quantify a gap: current LLMs perform limitedly on figure skating rules and event facts, so progress in artistic-sports knowledge is directly measurable.
  • Instruction-tuning on FSAnno improves captioning F1 from about 7% to about 38% under AutoDQ, indicating that fine-grained multi-modal annotations, not just more video data, drive the gain.
  • Because FSBench includes identity-free motion and skeleton data, it enables experiments on whether models can judge artistry without seeing the athlete's face or costume.
  • The annotation design is extensible to anomaly detection and motion generation tasks, which the paper lists as supported future uses.

Reading between the lines

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

  • Beyond the paper, the same three-source recipe—official score sheets, timestamp-aligned commentator audio, and LLM-synthesized summaries—looks directly transferable to other judged artistic sports such as gymnastics, dance, and synchronized swimming, where the missing ingredient is also multi-level technical plus artistic annotation.
  • Editorial inference: since the same kind of LLM used to synthesize training templates is also used to score model outputs through AutoDQ, part of SkateLLM's improvement may reflect stylistic alignment with the evaluator rather than deeper skating knowledge; the paper does not test this.
  • Editorial inference: a human study in which judges rate artistry from motion/skeleton data alone would test whether the identity-free representations preserve the artistic signal FSBench claims to evaluate.
  • Editorial inference: the paper's headline conclusion is currently demonstrated mainly on the captioning task; the claim about artistic understanding would be better supported if full results on assessment and commentary tasks appear in the released repository.
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 / 6 minor

Summary. The paper introduces FSAnno, a large-scale figure skating dataset with multi-level annotations spanning prior knowledge, individual actions, and whole performances, and FSBench, a benchmark for evaluating LLMs on figure skating understanding. The dataset is constructed from 783 competition performances with RGB, motion, skeleton, audio, and text modalities, and includes official judge reports as scoring ground truth. The authors also present SkateLLM, an instruction-tuned variant of Motion-GPT trained on FSAnno, and report that it substantially outperforms the base model on a motion-captioning task. The paper concludes that fine-grained multi-modal annotations significantly enhance LLM capabilities in artistic sports understanding.

Significance. If the annotations are validated, FSAnno/FSBench would fill a clear gap in sports-understanding benchmarks, which currently focus on ball sports and single-task figure skating recognition or scoring. The inclusion of multiple modalities, multiple task levels, and official judge reports is a strength, and the preliminary result that current LLMs score poorly on figure skating knowledge is an interesting finding. However, the central causal claim that fine-grained annotations 'significantly enhance' LLM capabilities is not established by the presented evidence, because the evaluation pipeline is circular and the empirical demonstration is limited to a single task. The benchmark has potential value, but the paper needs stronger validation and more thorough experiments before the claim can be accepted.

major comments (4)
  1. [§3.3–§3.4] Section 3.3 (Data Annotation) and Section 3.4: the entire annotation pipeline is automatic—4DHumans for motion extraction, HRNet for skeletons, Whisper for speech-to-text, and GPT-4 for generating summary comments—with no reported human validation, inter-annotator agreement, or error analysis. Because FSBench and FSAnno use these annotations as ground truth for all tasks (e.g., action segmentation aligned to video cue boxes, GOE scores, commentary), the reliability of the benchmark under this pipeline is not established. The authors should include a human verification study on a subset (with agreement statistics) or explicitly quantify the error modes of each automatic stage.
  2. [§3.5, §4.1, Table 4] Sections 3.5 and 4.1: the evaluation metric AutoDQ uses GPT-3.5-turbo for both event extraction and cross-checking, while the SkateLLM training captions are generated by GPT-4 from manually crafted templates (Section 4.1). Consequently, SkateLLM is trained to imitate the annotation pipeline's language, and AutoDQ measures similarity to that same language. The large F1 gain in Table 4 (7.1 to 38.0) may therefore reflect pipeline alignment rather than technically accurate understanding of figure skating; the claim in the abstract that annotations 'significantly enhance the LLMs' capabilities' requires either a human expert evaluation of the generated captions or an evaluation metric independent of the annotation generation process.
  3. [§5.2, Table 4] Section 5.2 and Table 4: the reported evidence for the enhancement claim is limited to a single task (element description) and a single baseline (Motion-GPT), with no error bars, no statistical testing, and no human evaluation. The paper itself acknowledges in Section 5.2 that existing video- and motion-based LLMs were only partially tested and that it 'focus[es] temporarily on the most fundamental task.' To support the abstract's general conclusion, results on additional FSBench tasks (e.g., action recognition, performance commentary) and comparisons with at least one video-based MLLM are needed.
  4. [Introduction vs. §3.3] Introduction (third contribution bullet) states that 'FSBench-Text includes multiple-choice questions with human-annotated explanations,' but Section 3.3 describes using a large language model to process collected commentary and does not describe any human annotation of the explanations. The authors should either correct the description or provide the annotation procedure for these explanations; this discrepancy currently undermines the paper's claim of fine-grained human-quality annotations.
minor comments (6)
  1. [Table 3] The model names are inconsistent (e.g., 'GPT3.5-turbo', 'GPT4', 'LLaV A 13B'); please standardize the notation.
  2. [§5.1] The '5 shot strategy' is not described; please clarify how the few-shot examples are selected.
  3. [§4.2] Training details are too sparse (e.g., learning rate, batch size, optimizer, number of steps) to reproduce SkateLLM; please provide these hyperparameters.
  4. [Figure 2] The pie chart and stacked bar chart are small and difficult to read; please increase the font sizes.
  5. [Table 2] The category abbreviations (e.g., 'LS' for long-video scoring) are not defined in the caption; please add a full list of abbreviations.
  6. [Abstract/§3.3] The paper claims that FSAnno is 'open-access' and points to an official repository, but no download link is provided anywhere in the manuscript; please add the link in the final version.

Circularity Check

2 steps flagged · score 5.0 of 10

Reported SkateLLM gain is a self-consistency score between GPT-4-generated training captions and a GPT-3.5-based metric extracting the same template events.

  1. self definitional [§4.1 (Instruction-Tuning Data) and §3.5 (Evaluation Method)]
    "We use GPT4 to generate data in batches according to the template. Specifically, we used GPT-4 to generate descriptions for different element categories, creating templates that included technical details like take-off edge, rotation, air position, and landing control. ... In the first stage, key events are extracted from the ground-truth description and the LLM-generated description. These events represent core elements of figure skating movements, such as the edge of take-off, rotation direction, aerial posture, and landing details."

    The training captions are written by GPT-4 from templates whose content slots are 'take-off edge, rotation, air position, and landing control.' The evaluation metric AutoDQ defines 'key events' as 'the edge of take-off, rotation direction, aerial posture, and landing details' — the same content categories. A model that imitates the GPT-4 template language will have its extracted events matched by AutoDQ and receive a higher F1. The metric is not an external check of whether the caption is true of the observed motion; it checks alignment with the same annotation vocabulary used to build the training target. The improvement is therefore partly a self-consistency score between the training generator and the evaluation generator, not an independent demonstration of enhanced understanding.

  2. fitted input called prediction [§5.2, Table 4 and Figure 5]
    "Table 4 shows the results of different motion-based LLMs evaluated using AutoDQ. It can be observed that the performance of Motion-GPT, which has not undergone instruction fine-tuning, is unsatisfactory despite being trained on various motion datasets. SkateLLM exhibits higher precision but lower recall... For event extraction and cross-checking, we use GPT-3.5-turbo."

    The only quantitative support for the central claim that FSAnno annotations 'significantly enhance the LLMs' capabilities' is Table 4's F1 gain (7.1 to 38.0) on element description, computed with AutoDQ. The ground-truth key events come from the same FSAnno/GPT-4 generation pipeline used to produce SkateLLM's instruction-tuning targets, and the metric itself is another LLM (GPT-3.5-turbo) performing event extraction. A model can raise AutoDQ by reproducing template-style event phrasing rather than by technically accurate description of the actual skating elements. No human evaluation or error analysis is provided to break this loop, so the measured improvement is not an independent prediction but a match to the annotation distribution it was fitted on.

full rationale

The paper's dataset construction and prior-knowledge tests are not circular: FSBench-Text is described as containing human-annotated multiple-choice questions, and Table 3 is compared against an external quiz classification. No load-bearing self-citation or imported uniqueness theorem appears. The circularity is confined to the SkateLLM enhancement claim. The training captions in §4.1 are generated by GPT-4 from manually crafted templates, while the evaluation in §3.5 uses AutoDQ, an LLM-based metric that extracts 'key events' from both the ground truth and the generated caption. The event vocabulary in the evaluation is the same as the template vocabulary in training, and the ground-truth captions themselves come from the same automated annotation pipeline, with no reported human validation or error analysis. Consequently, the reported F1 improvement may reflect the model learning to mimic GPT-4's annotation style rather than genuinely improved figure-skating understanding. This is a partial circularity affecting the causal claim, while the benchmark itself may still be a useful resource. Score 5 reflects that the central enhancement claim is substantially undermined by the self-referential training/evaluation loop, but the prior-knowledge benchmark and dataset contributions retain independent content.

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

The central claims rest primarily on the accuracy of the automatic annotation pipeline and the validity of the LLM-based evaluation metric; no free parameters or invented physical entities are introduced. The paper would be strengthened by human validation of these assumptions.

assumptions (3)
  • domain assumption Automatic extraction of motion and skeleton data (4DHumans, HRNet) faithfully represents the athletic and artistic content of figure skating performances.
    Section 3.3: These modalities are the primary inputs for FSBench-Motion tasks; no validation against manual annotation or human expert judgment is provided.
  • ad hoc to paper The LLM-based metric AutoDQ reliably measures semantic alignment between generated and ground-truth descriptions of figure skating elements.
    Section 3.5: AutoDQ uses GPT-3.5-turbo for event extraction and matching; no human evaluation or correlation study with human judgments is reported.
  • domain assumption Official judge reports and Whisper-transcribed commentator audio provide accurate ground truth for element categories, scores, and commentary.
    Section 3.3: The annotations are derived from these sources; although judge reports are authoritative, the speech-to-text step and the timestamp-based assignment of comments to actions are not error-checked.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FSBench: A Figure Skating Benchmark for Advancing Artistic Sports Understanding." pith.science (2026). https://pith.science/paper/DOM3GBWU

@misc{pith2026250419514,
  author       = {Pith},
  title        = {Pith review of: FSBench: A Figure Skating Benchmark for Advancing Artistic Sports Understanding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DOM3GBWU}},
  note         = {Machine review of arXiv:2504.19514}
}
read the original abstract

Figure skating, known as the "Art on Ice," is among the most artistic sports, challenging to understand due to its blend of technical elements (like jumps and spins) and overall artistic expression. Existing figure skating datasets mainly focus on single tasks, such as action recognition or scoring, lacking comprehensive annotations for both technical and artistic evaluation. Current sports research is largely centered on ball games, with limited relevance to artistic sports like figure skating. To address this, we introduce FSAnno, a large-scale dataset advancing artistic sports understanding through figure skating. FSAnno includes an open-access training and test dataset, alongside a benchmark dataset, FSBench, for fair model evaluation. FSBench consists of FSBench-Text, with multiple-choice questions and explanations, and FSBench-Motion, containing multimodal data and Question and Answer (QA) pairs, supporting tasks from technical analysis to performance commentary. Initial tests on FSBench reveal significant limitations in existing models' understanding of artistic sports. We hope FSBench will become a key tool for evaluating and enhancing model comprehension of figure skating.

Figures

Figures reproduced from arXiv: 2504.19514 by the authors.

Figure 1
Figure 1. To systematically assess whether models understand figure skating which is an artistic sport, FSBench sources its data from [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Data Statistics. (a) The figure skating elements are divided into three main categories: Jump, Spin, and Sequence, which are [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Our annotations are derived from three sources: official judging reports, visual information, and audio commentary. These three [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Compared to other large language models, SkateLLM’s [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 27 canonical work pages

  1. [1]

    Gpt-4 technical report

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023. 2

    arXiv 2023

  2. [2]

    Meteor: An auto- matic metric for mt evaluation with improved correla- tion with human judgments

    Satanjeev Banerjee and Alon Lavie. Meteor: An auto- matic metric for mt evaluation with improved correla- tion with human judgments. In Proceedings of the acl workshop on intrinsic and extrinsic evaluation mea- sures for machine translation and/or summarization , pages 65–72, 2005. 6

    2005

  3. [3]

    Mo- tionllm: Understanding human behaviors from human motions and videos.arXiv preprint arXiv:2405.20340,

    Ling-Hao Chen, Shunlin Lu, Ailing Zeng, Hao Zhang, Benyou Wang, Ruimao Zhang, and Lei Zhang. Mo- tionllm: Understanding human behaviors from human motions and videos.arXiv preprint arXiv:2405.20340,

    arXiv

  4. [4]

    Machine and deep learning for sport- specific movement recognition: A systematic review of model development and performance

    Emily E Cust, Alice J Sweeting, Kevin Ball, and Sam Robertson. Machine and deep learning for sport- specific movement recognition: A systematic review of model development and performance. Journal of sports sciences, 37(5):568–600, 2019. 2

    work page 2019

  5. [5]

    Soccernet-v2: A dataset and benchmarks for holistic understanding of broad- cast soccer videos

    Adrien Deliege, Anthony Cioppa, Silvio Giancola, Meisam J Seikavandi, Jacob V Dueholm, Kamal Nas- rollahi, Bernard Ghanem, Thomas B Moeslund, and Marc Van Droogenbroeck. Soccernet-v2: A dataset and benchmarks for holistic understanding of broad- cast soccer videos. In Proceedings of the IEEE/CVF conference on computer vision and pattern recogni- tion, pag...

    work page 2021

  6. [6]

    Learn- ing semantics-guided representations for scoring fig- ure skating

    Zexing Du, Di He, Xue Wang, and Qing Wang. Learn- ing semantics-guided representations for scoring fig- ure skating. IEEE Transactions on Multimedia, 2023. 1, 3, 5

    work page 2023

  7. [7]

    Videoagent: A memory- augmented multimodal agent for video understanding

    Yue Fan, Xiaojian Ma, Rujie Wu, Yuntao Du, Jiaqi Li, Zhi Gao, and Qing Li. Videoagent: A memory- augmented multimodal agent for video understanding. In European Conference on Computer Vision , pages 75–92. Springer, 2025. 2

    work page 2025

  8. [8]

    Soccernet: A scalable dataset for action spotting in soccer videos

    Silvio Giancola, Mohieddine Amine, Tarek Dghaily, and Bernard Ghanem. Soccernet: A scalable dataset for action spotting in soccer videos. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops, pages 1711–1721, 2018. 2

    work page 2018

Show all 46 references
  1. [9]

    Hu- mans in 4d: Reconstructing and tracking humans with transformers

    Shubham Goel, Georgios Pavlakos, Jathushan Ra- jasegaran, Angjoo Kanazawa, and Jitendra Malik. Hu- mans in 4d: Reconstructing and tracking humans with transformers. In Proceedings of the IEEE/CVF In- ternational Conference on Computer Vision , pages 14783–14794, 2023. 5

    2023

  2. [10]

    Generating diverse and natural 3d human motions from text

    Chuan Guo, Shihao Zou, Xinxin Zuo, Sen Wang, Wei Ji, Xingyu Li, and Li Cheng. Generating diverse and natural 3d human motions from text. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5152–5161, 2022. 7

    2022

  3. [11]

    A perennial dilemma: Artistic sports in the olympic games

    Elizabeth A Hanley. A perennial dilemma: Artistic sports in the olympic games. Journal of Olympic His- tory, 8(2):39–46, 2000. 1

    2000

  4. [12]

    Effi- cient multimodal learning from data-centric perspec- tive

    Muyang He, Yexin Liu, Boya Wu, Jianhao Yuan, Yueze Wang, Tiejun Huang, and Bo Zhao. Effi- cient multimodal learning from data-centric perspec- tive. arXiv preprint arXiv:2402.11530, 2024. 7

    2024 arXiv

  5. [13]

    Lora: Low-rank adaptation of large language models

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685 ,

    arXiv

  6. [14]

    Gpt-4o system card

    Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Os- trow, Akila Welihinda, Alan Hayes, Alec Rad- ford, et al. Gpt-4o system card. arXiv preprint arXiv:2410.21276, 2024. 2

    2024 arXiv

  7. [15]

    Motiongpt: Human motion as a foreign language

    Biao Jiang, Xin Chen, Wen Liu, Jingyi Yu, Gang Yu, and Tao Chen. Motiongpt: Human motion as a foreign language. Advances in Neural Information Processing Systems, 36:20067–20079, 2023. 3, 8

    2023

  8. [16]

    Chat-univi: Unified visual representation empowers large language models with image and video understanding

    Peng Jin, Ryuichi Takanobu, Wancai Zhang, Xi- aochun Cao, and Li Yuan. Chat-univi: Unified visual representation empowers large language models with image and video understanding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 137...

    2024

  9. [17]

    Seed- bench: Benchmarking multimodal large language models

    Bohao Li, Yuying Ge, Yixiao Ge, Guangzhi Wang, Rui Wang, Ruimao Zhang, and Ying Shan. Seed- bench: Benchmarking multimodal large language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 13299–13308, 2024. 3

    2024

  10. [18]

    Sports-qa: A large-scale video question answering benchmark for complex and professional sports

    Haopeng Li, Andong Deng, Qiuhong Ke, Jun Liu, Hossein Rahmani, Yulan Guo, Bernt Schiele, and Chen Chen. Sports-qa: A large-scale video question answering benchmark for complex and professional sports. arXiv preprint arXiv:2401.01505, 2024. 2, 3

    2024

  11. [19]

    Mvbench: A comprehensive multi-modal video understanding benchmark

    Kunchang Li, Yali Wang, Yinan He, Yizhuo Li, Yi Wang, Yi Liu, Zun Wang, Jilan Xu, Guo Chen, Ping Luo, et al. Mvbench: A comprehensive multi-modal video understanding benchmark. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 22195–22...

    2024

  12. [20]

    Llama-vid: An image is worth 2 tokens in large language models

    Yanwei Li, Chengyao Wang, and Jiaya Jia. Llama-vid: An image is worth 2 tokens in large language models. In European Conference on Computer Vision , pages 323–340. Springer, 2025. 2

    2025

  13. [21]

    Video-llava: Learning united 9 visual representation by alignment before projection

    Bin Lin, Yang Ye, Bin Zhu, Jiaxi Cui, Munan Ning, Peng Jin, and Li Yuan. Video-llava: Learning united 9 visual representation by alignment before projection. arXiv preprint arXiv:2311.10122, 2023. 2

    2023 arXiv

  14. [22]

    Liveqa: A question answering dataset over sports live

    Qianying Liu, Sicong Jiang, Yizhong Wang, and Su- jian Li. Liveqa: A question answering dataset over sports live. In Chinese Computational Linguistics: 19th China National Conference, CCL 2020, Hainan, China, October 30–November 1, 2020, Proceedings 19, pages 316–328. Springer...

    2020

  15. [23]

    Fsd-10: A fine-grained classification dataset for figure skating

    Shenglan Liu, Xiang Liu, Gao Huang, Hong Qiao, Lianyu Hu, Dong Jiang, Aibin Zhang, Yang Liu, and Ge Guo. Fsd-10: A fine-grained classification dataset for figure skating. Neurocomputing, 413:360–367,

  16. [24]

    Tempo- ral segmentation of fine-gained semantic action: A motion-centered figure skating dataset

    Shenglan Liu, Aibin Zhang, Yunheng Li, Jian Zhou, Li Xu, Zhuben Dong, and Renhao Zhang. Tempo- ral segmentation of fine-gained semantic action: A motion-centered figure skating dataset. In Proceed- ings of the AAAI conference on artificial intelligence , pages 2163–2171, 2021....

    2021

  17. [25]

    Fine-grained action analysis: A multi-modality and multi-task dataset of figure skating

    Sheng-Lan Liu, Yu-Ning Ding, Si-Fan Zhang, Wen- Yue Chen, Ning Zhou, Hao Liu, and Gui-Hong Lao. Fine-grained action analysis: A multi-modality and multi-task dataset of figure skating. arXiv preprint arXiv:2307.02730, 2023. 1

    2023 arXiv

  18. [26]

    Gleu: Automatic evaluation of sentence-level fluency

    Andrew Mutton, Mark Dras, Stephen Wan, and Robert Dale. Gleu: Automatic evaluation of sentence-level fluency. In Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics , pages 344–351, 2007. 6

    2007

  19. [27]

    Bleu: a method for automatic evalua- tion of machine translation

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. Bleu: a method for automatic evalua- tion of machine translation. InProceedings of the 40th annual meeting of the Association for Computational Linguistics, pages 311–318, 2002. 6

    2002

  20. [28]

    Assessing the quality of actions

    Hamed Pirsiavash, Carl V ondrick, and Antonio Tor- ralba. Assessing the quality of actions. In Com- puter Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceed- ings, Part VI 13 , pages 556–571. Springer, 2014. 1, 3

    2014

  21. [29]

    The kit motion-language dataset

    Matthias Plappert, Christian Mandery, and Tamim As- four. The kit motion-language dataset. Big data, 4(4): 236–252, 2016. 7

    2016

  22. [30]

    Ro- bust speech recognition via large-scale weak supervi- sion

    Alec Radford, Jong Wook Kim, Tao Xu, Greg Brock- man, Christine McLeavey, and Ilya Sutskever. Ro- bust speech recognition via large-scale weak supervi- sion. In International conference on machine learn- ing, pages 28492–28518. PMLR, 2023. 6

    2023

  23. [31]

    A survey of content-aware video analysis for sports

    Huang-Chia Shih. A survey of content-aware video analysis for sports. IEEE Transactions on circuits and systems for video technology, 28(5):1212–1231, 2017. 2

    2017

  24. [32]

    Beyond the imitation game: Quantifying and extrapolating the capabilities of lan- guage models

    Aarohi Srivastava, Abhinav Rastogi, Abhishek Rao, Abu Awal Md Shoeb, Abubakar Abid, Adam Fisch, Adam R Brown, Adam Santoro, Aditya Gupta, Adri `a Garriga-Alonso, et al. Beyond the imitation game: Quantifying and extrapolating the capabilities of lan- guage models. arXiv prepri...

    arXiv

  25. [33]

    Deep high-resolution representation learning for hu- man pose estimation

    Ke Sun, Bin Xiao, Dong Liu, and Jingdong Wang. Deep high-resolution representation learning for hu- man pose estimation. In CVPR, 2019. 5

    2019

  26. [34]

    Tarsier: Recipes for training and evalu- ating large video description models

    Jiawei Wang, Liping Yuan, Yuchen Zhang, and Hao- miao Sun. Tarsier: Recipes for training and evalu- ating large video description models. arXiv preprint arXiv:2407.00634, 2024. 6

    2024 arXiv

  27. [35]

    Motion-agent: A con- versational framework for human motion generation with llms

    Qi Wu, Yubo Zhao, Yifan Wang, Xinhang Liu, Yu- Wing Tai, and Chi-Keung Tang. Motion-agent: A con- versational framework for human motion generation with llms. arXiv preprint arXiv:2405.17013, 2024. 3

    2024 arXiv

  28. [36]

    Sportqa: A benchmark for sports understanding in large language models

    Haotian Xia, Zhengbang Yang, Yuqing Wang, Rhys Tracy, Yun Zhao, Dongdong Huang, Zezhi Chen, Yan Zhu, Yuan-fang Wang, and Weining Shen. Sportqa: A benchmark for sports understanding in large language models. arXiv preprint arXiv:2402.15862, 2024. 2, 3

    2024 arXiv

  29. [37]

    Sportu: A comprehensive sports understanding benchmark for multimodal large language models

    Haotian Xia, Zhengbang Yang, Junbo Zou, Rhys Tracy, Yuqing Wang, Chi Lu, Christopher Lai, Yan- jun He, Xun Shao, Zhuoqing Xie, et al. Sportu: A comprehensive sports understanding benchmark for multimodal large language models. arXiv preprint arXiv:2410.08474, 2024. 2, 3

    arXiv

  30. [38]

    Skating-mixer: Long-term sport audio-visual model- ing with mlps

    Jingfei Xia, Mingchen Zhuge, Tiantian Geng, Shun Fan, Yuantai Wei, Zhenyu He, and Feng Zheng. Skating-mixer: Long-term sport audio-visual model- ing with mlps. In Proceedings of the AAAI Conference on Artificial Intelligence, pages 2901–2909, 2023. 1, 3, 5

    2023

  31. [39]

    Simple base- lines for human pose estimation and tracking

    Bin Xiao, Haiping Wu, and Yichen Wei. Simple base- lines for human pose estimation and tracking. InEuro- pean Conference on Computer Vision (ECCV) , 2018. 5

    2018

  32. [40]

    Learning to score figure skating sport videos

    Chengming Xu, Yanwei Fu, Bing Zhang, Zitian Chen, Yu-Gang Jiang, and Xiangyang Xue. Learning to score figure skating sport videos. IEEE transactions on circuits and systems for video technology , 30(12): 4578–4590, 2019. 1, 3

    2019

  33. [41]

    Spatio- temporal dynamic inference network for group ac- tivity recognition

    Hangjie Yuan, Dong Ni, and Mang Wang. Spatio- temporal dynamic inference network for group ac- tivity recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 7476–7485, 2021. 2

    2021

  34. [42]

    Video- llama: An instruction-tuned audio-visual language model for video understanding

    Hang Zhang, Xin Li, and Lidong Bing. Video- llama: An instruction-tuned audio-visual language model for video understanding. arXiv preprint arXiv:2306.02858, 2023. 2 10

    2023 arXiv

  35. [43]

    Flash- vstream: Memory-based real-time understanding for long video streams

    Haoji Zhang, Yiqin Wang, Yansong Tang, Yong Liu, Jiashi Feng, Jifeng Dai, and Xiaojie Jin. Flash- vstream: Memory-based real-time understanding for long video streams. arXiv preprint arXiv:2406.08085,

    arXiv

  36. [44]

    A sur- vey of deep learning in sports applications: Percep- tion, comprehension, and decision

    Zhonghan Zhao, Wenhao Chai, Shengyu Hao, Wen- hao Hu, Guanhong Wang, Shidong Cao, Mingli Song, Jenq-Neng Hwang, and Gaoang Wang. A sur- vey of deep learning in sports applications: Percep- tion, comprehension, and decision. arXiv preprint arXiv:2307.03353, 2023. 2

    2023 arXiv

  37. [45]

    Avatargpt: All-in-one framework for motion under- standing planning generation and beyond

    Zixiang Zhou, Yu Wan, and Baoyuan Wang. Avatargpt: All-in-one framework for motion under- standing planning generation and beyond. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1357–1366, 2024. 3

    2024

  38. [46]

    Mme: A comprehensive evaluation benchmark for mul- timodal large language models

    Qingyang Zhu, Shaohua Chen, Zhengfu Liu, Wendi Zhou, Ziyi Wang, Jindong Wang, Baosheng Shi, Rui Zhang, Yongdong Zhang, and Nan Lu. Mme: A comprehensive evaluation benchmark for mul- timodal large language models. arXiv preprint arXiv:2306.13394, 2023. 3 11

    2023 arXiv

Pith tools

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