pith. sign in

arxiv: 2605.11208 · v2 · pith:Q5BPFQLCnew · submitted 2026-05-11 · 💻 cs.CV

Hi-GaTA: Hierarchical Gated Temporal Aggregation Adapter for Surgical Video Report Generation

Kedi Sun , Chaohui Dang , Yue Feng , James Glasbey , Theodoros N. Arvanitis , Le Zhang This is my paper

Pith reviewed 2026-05-20 21:43 UTC · model grok-4.3

classification 💻 cs.CV
keywords surgical video report generationtemporal aggregation adaptermultimodal large language modelvideo encoder pretraininghierarchical gated fusionLoRA fine-tuningsimulated surgical benchmark
0
0 comments X

The pith

Hi-GaTA compresses long surgical videos into compact LLM tokens via hierarchical gated temporal aggregation to generate clinician-style reports.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a benchmark of 214 simulated surgical videos paired with surgeon-authored reports to support automated report generation. It proposes a Perception-Alignment-Reasoning framework that includes a pretrained surgical video encoder and the Hi-GaTA adapter, which aggregates temporal features at multiple scales using text-conditioned cross-attention and gated fusion before feeding prefix tokens to a fine-tuned LLM. This setup aims to align dense video representations with language reasoning while using limited supervision through LoRA. Experiments demonstrate consistent gains over strong multimodal LLM baselines on the benchmark. The work targets reduced documentation burden in surgery through objective, automated feedback.

Core claim

The central claim is that a lightweight hierarchical gated temporal aggregation adapter, when combined with a domain-specific video encoder pretrained on 40,000 minutes of surgical footage, produces superior surgical video reports by efficiently compressing spatio-temporal sequences into LLM-compatible tokens through short-to-long range aggregation, dual cross-attention, cross-level gated fusion, and an increasing-depth strategy.

What carries the argument

Hi-GaTA, a hierarchical gated temporal aggregation adapter that employs a temporal pyramid with text-conditioned dual cross-attention, cross-level gated fusion, and increasing-depth processing to compress video sequences into compact visual prefix tokens.

If this is right

  • Automated generation of clinician-grade assessment reports becomes feasible for surgical procedures.
  • Documentation burden decreases while objective feedback on procedural quality increases.
  • The approach yields the best overall performance with consistent improvements over multimodal LLM baselines on the established benchmark.
  • Each component of the adapter, including the gated fusion and multi-scale consistency mechanisms, contributes measurably to the results as shown by ablations.
  • Fine-tuning via LoRA enables coherent and stylistically consistent report generation under limited supervision.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the benchmark patterns generalize, the adapter could support real-time procedural feedback during live surgeries.
  • The pretraining on large-scale public surgical videos suggests the encoder could serve as a reusable backbone for other video-based surgical analysis tasks.
  • Extending the temporal pyramid to even longer procedures might require additional memory-efficient variants of the gated fusion step.
  • Pairing the generated reports with outcome data could enable downstream models that predict complication risks from video alone.

Load-bearing premise

The simulated surgical videos and surgeon-authored reports in the 214-video benchmark capture the spatio-temporal patterns and linguistic style of real clinical procedures so that benchmark gains transfer to actual use.

What would settle it

A direct comparison showing that the model produces substantially lower-quality or less accurate reports when tested on a collection of real clinical surgical videos paired with actual surgeon reports would indicate the simulated benchmark does not support transfer.

Figures

Figures reproduced from arXiv: 2605.11208 by Chaohui Dang, James Glasbey, Kedi Sun, Le Zhang, Theodoros N. Arvanitis, Yue Feng.

Figure 1
Figure 1. Figure 1: Overview of our proposed method. Left: The Perception–Alignment–Reasoning pipeline for surgical video report generation. Right: Detailed architecture of the Hi￾GaTA module. linear layers with GELU activation and LayerNorm). The resulting projections, z (1) , z (2) ∈ R D are ℓ2-normalized. We optimize the encoder using a symmetric InfoNCE objective [16], which treats (z (1) i , z (2) i ) as a positive pair … view at source ↗
Figure 2
Figure 2. Figure 2: Qualitative comparison of generated reports. Our Hi-GaTA approach produces more clinically accurate and comprehensive descriptions than LLaVA-Med-v1.5-7B [8] and Qwen2.5-VL-7B [2], closer to the ground truth. when paired with Sur40k. The large CIDEr variance stems from the high lin￾guistic variability of expert narratives and strict n-gram matching, which heavily penalizes clinically valid synonymous descr… view at source ↗
read the original abstract

Automated, clinician-grade assessment reports for surgical procedures could reduce documentation burden and provide objective feedback, yet remain challenging due to the difficulty of aligning dense spatio-temporal video representations with language-based reasoning and the scarcity of high-quality, privacy-preserving datasets. To address this gap, we establish a benchmark comprising 214 high-quality simulated surgical videos paired with surgeon-authored evaluation reports. Building on this resource, we propose a Perception-Alignment-Reasoning framework for surgical video report generation, featuring Hi-GaTA, a novel lightweight temporal adapter that efficiently compresses long video sequences into compact, LLM-compatible visual prefix tokens through short-to-long-range temporal aggregation. For robust visual perception, we pretrain Sur40k, a surgical-specific ViViT-style video encoder on 40,000 minutes of public surgical videos to capture fine-grained spatio-temporal procedural priors. Hi-GaTA employs a temporal pyramid with text-conditioned dual cross-attention, and improves multi-scale consistency through cross-level gated fusion and an increasing-depth strategy. Finally, we fine-tune the LLM backbone using LoRA to enable coherent and stylistically consistent surgical report generation under limited supervision. Experiments show our approach achieves the best overall performance, with consistent gains over strong Multimodal Large Language Model (MLLM) baselines. Ablation studies further validate the effectiveness of each proposed component.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript establishes a benchmark of 214 simulated surgical videos paired with surgeon-authored reports and proposes a Perception-Alignment-Reasoning framework for automated surgical video report generation. Central to the approach is Hi-GaTA, a lightweight hierarchical gated temporal aggregation adapter that compresses long video sequences into LLM-compatible tokens via a temporal pyramid, text-conditioned dual cross-attention, cross-level gated fusion, and increasing-depth strategy. The framework includes pretraining a surgical-specific ViViT-style encoder (Sur40k) on 40,000 minutes of public surgical videos, followed by LoRA fine-tuning of the LLM backbone. The central claim is that this yields the best overall performance with consistent gains over strong MLLM baselines, supported by ablation studies on component effectiveness.

Significance. If the reported gains hold and prove transferable, the work could meaningfully reduce documentation burden in surgical settings by enabling efficient, temporally-aware report generation from video. The creation of a surgeon-authored benchmark and the domain-specific pretraining of Sur40k on a large public corpus are concrete contributions that facilitate future research in surgical video understanding. The hierarchical gated adapter design offers a practical mechanism for handling long spatio-temporal sequences without full fine-tuning. However, the overall significance is limited by the exclusive use of simulated data, which leaves open whether the improvements reflect genuine advances in modeling real clinical variability.

major comments (2)
  1. [Abstract and Benchmark Construction] The headline claim that the approach achieves the best overall performance with gains over MLLM baselines rests on results from the 214-video simulated benchmark. The manuscript provides no quantitative domain-shift metrics (e.g., distribution distances on visual features, camera motion statistics, or tissue deformation patterns) or side-by-side comparisons of simulated versus real surgical reports, making it impossible to determine whether the observed improvements would transfer to real OR footage. This is load-bearing for the central claim of a general advance in surgical report generation.
  2. [Experiments and Ablations] Ablation studies are cited to validate each component of Hi-GaTA, yet the abstract and description supply no numerical results, error bars, or baseline details for these ablations. Without these numbers it is difficult to assess whether the temporal pyramid, gated fusion, or increasing-depth strategy contribute meaningfully beyond standard adapter baselines.
minor comments (2)
  1. [Method] The notation and exact formulation of the text-conditioned dual cross-attention and cross-level gated fusion would benefit from an explicit equation or pseudocode block to improve reproducibility.
  2. [Related Work] The paper would be strengthened by citing prior work on surgical video analysis and existing video-to-text benchmarks to better situate the novelty of the 214-video resource.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the thoughtful and constructive review. We address the two major comments point by point below, indicating where revisions will be made to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract and Benchmark Construction] The headline claim that the approach achieves the best overall performance with gains over MLLM baselines rests on results from the 214-video simulated benchmark. The manuscript provides no quantitative domain-shift metrics (e.g., distribution distances on visual features, camera motion statistics, or tissue deformation patterns) or side-by-side comparisons of simulated versus real surgical reports, making it impossible to determine whether the observed improvements would transfer to real OR footage. This is load-bearing for the central claim of a general advance in surgical report generation.

    Authors: We agree that the absence of explicit domain-shift quantification and real-data comparisons limits the strength of claims about transfer to clinical OR settings. Our benchmark is intentionally built on high-quality simulated videos with surgeon-authored reports to address privacy and data-scarcity barriers that currently prevent large-scale real OR datasets. In the revised manuscript we will (1) add quantitative domain-gap analysis using available public real surgical video clips (e.g., feature-distribution distances and motion statistics between our simulated set and subsets of public real footage), (2) expand the limitations section to explicitly discuss the simulated-to-real gap, and (3) temper the abstract and conclusion to state that performance gains are demonstrated on the proposed simulated benchmark while noting the need for future real-data validation. We cannot provide paired real surgical reports at this time due to ethical and privacy constraints. revision: partial

  2. Referee: [Experiments and Ablations] Ablation studies are cited to validate each component of Hi-GaTA, yet the abstract and description supply no numerical results, error bars, or baseline details for these ablations. Without these numbers it is difficult to assess whether the temporal pyramid, gated fusion, or increasing-depth strategy contribute meaningfully beyond standard adapter baselines.

    Authors: We apologize for the lack of explicit numerical summaries in the abstract and high-level description. The full paper already contains detailed ablation tables (Section 4.3) that report exact metrics, standard deviations across runs, and comparisons against standard adapter baselines for each Hi-GaTA component. In the revision we will insert a concise summary of the key ablation results (including numerical deltas and error bars) directly into the abstract and add explicit cross-references in the method section so readers can immediately locate the supporting numbers without searching the supplementary material. revision: yes

standing simulated objections not resolved
  • Direct side-by-side evaluation on real operating-room footage with paired surgeon reports, which remains unavailable due to privacy regulations and ethical restrictions on clinical video collection.

Circularity Check

0 steps flagged

No circularity: standard pretrain-adapt pipeline with independent empirical validation

full rationale

The paper presents an architectural proposal (Hi-GaTA temporal adapter with pyramid, dual cross-attention, gated fusion, and increasing-depth strategy) inside a Perception-Alignment-Reasoning framework, pretrained on external public surgical video data (Sur40k) and evaluated on a new 214-video simulated benchmark. No equations, first-principles derivations, or predictions are offered that reduce by construction to fitted parameters or self-definitions. Ablations and baseline comparisons constitute independent empirical content rather than tautological renaming or self-citation load-bearing. The central claims rest on observed performance gains, not on any step that is definitionally equivalent to its inputs.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 2 invented entities

The central claims rest on the new model components and the assumption that simulated data plus public surgical video pretraining provide sufficient priors; no external machine-checked proofs or parameter-free derivations are referenced.

free parameters (2)
  • Hi-GaTA pyramid scales and gating thresholds
    Hyperparameters chosen to control multi-scale temporal aggregation and fusion.
  • LoRA rank and surgical-specific fine-tuning hyperparameters
    Standard adaptation parameters fitted during limited-supervision training.
axioms (1)
  • domain assumption Simulated videos and surgeon-authored reports capture the essential spatio-temporal and linguistic characteristics needed for real surgical report generation
    Invoked when establishing the 214-video benchmark as a proxy for clinical use.
invented entities (2)
  • Hi-GaTA adapter no independent evidence
    purpose: Compress long video sequences into compact LLM-compatible visual prefix tokens via hierarchical temporal aggregation
    New model component introduced to solve the alignment problem between dense video and language reasoning.
  • Sur40k surgical video encoder no independent evidence
    purpose: Capture fine-grained spatio-temporal procedural priors from 40,000 minutes of public surgical videos
    New pretrained ViViT-style backbone specific to the surgical domain.

pith-pipeline@v0.9.0 · 5783 in / 1358 out tokens · 50609 ms · 2026-05-20T21:43:33.333219+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

31 extracted references · 31 canonical work pages · 5 internal anchors

  1. [1]

    In: Proceedings of the IEEE/CVF international confer- ence on computer vision

    Arnab, A., Dehghani, M., Heigold, G., Sun, C., Lučić, M., Schmid, C.: Vivit: A video vision transformer. In: Proceedings of the IEEE/CVF international confer- ence on computer vision. pp. 6836–6846 (2021)

    work page 2021

  2. [2]

    Qwen2.5-VL Technical Report

    Bai, S., Chen, K., Liu, X., Wang, J., Ge, W., Song, S., Dang, K., Wang, P., Wang, S., Tang, J., et al.: Qwen2. 5-vl technical report. arXiv preprint arXiv:2502.13923 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  3. [3]

    In: Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summa- rization

    Banerjee, S., Lavie, A.: Meteor: An automatic metric for mt evaluation with im- proved correlation with human judgments. In: Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summa- rization. pp. 65–72 (2005)

    work page 2005

  4. [4]

    arXiv e-prints pp

    Dubey, A., Jauhri, A., Pandey, A., Kadian, A., Al-Dahle, A., Letman, A., Mathur, A., Schelten, A., Yang, A., Fan, A., et al.: The llama 3 herd of models. arXiv e-prints pp. arXiv–2407 (2024)

    work page 2024

  5. [5]

    ACM Transactions on Computing for Healthcare (HEALTH) 3(1), 1–23 (2021)

    Gu, Y., Tinn, R., Cheng, H., Lucas, M., Usuyama, N., Liu, X., Naumann, T., Gao, J., Poon, H.: Domain-specific language model pretraining for biomedical natural language processing. ACM Transactions on Computing for Healthcare (HEALTH) 3(1), 1–23 (2021)

    work page 2021

  6. [6]

    ICLR1(2), 3 (2022)

    Hu, E.J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., Chen, W., et al.: Lora: Low-rank adaptation of large language models. ICLR1(2), 3 (2022)

    work page 2022

  7. [7]

    Medical Image Analysis p

    de Jong, R., Carolus, H., Franciscus, H., van Jaarsveld, R.C., van Hillegersberg, R., Josien, P., de With, P.H., al Khalil, Y., van Der Sommen, F., et al.: Scaling up self-supervised learning for improved surgical foundation models. Medical Image Analysis p. 103873 (2025)

    work page 2025

  8. [8]

    Advances in Neural Information Processing Systems36, 28541–28564 (2023)

    Li, C., Wong, C., Zhang, S., Usuyama, N., Liu, H., Yang, J., Naumann, T., Poon, H., Gao, J.: Llava-med: Training a large language-and-vision assistant for biomedicine in one day. Advances in Neural Information Processing Systems36, 28541–28564 (2023)

    work page 2023

  9. [9]

    In: International conference on machine learning

    Li, J., Li, D., Savarese, S., Hoi, S.: Blip-2: Bootstrapping language-image pre- training with frozen image encoders and large language models. In: International conference on machine learning. pp. 19730–19742. PMLR (2023)

    work page 2023

  10. [10]

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

    Li, K., Wang, Y., He, Y., Li, Y., Wang, Y., Liu, Y., Wang, Z., Xu, J., Chen, G., Luo, P., et al.: Mvbench: A comprehensive multi-modal video understanding benchmark. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 22195–22206 (2024)

    work page 2024

  11. [11]

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

    Li, S., Qin, P., Wu, H., Nie, D., Thirunavukarasu, A.J., Yu, J., Zhang, L.:µ2 tok- enizer: Differentiable multi-scale multi-modal tokenizer for radiology report gener- ation. In: International Conference on Medical Image Computing and Computer- Assisted Intervention. pp. 3–12. Springer (2025)

    work page 2025

  12. [12]

    arXiv preprint arXiv:2410.08588 (2024) 10 K

    Li, S., Xu, B., Luo, Y., Nie, D., Zhang, L.: Vit3d alignment of llama3: 3d medical image report generation. arXiv preprint arXiv:2410.08588 (2024) 10 K. Sun et al

    work page arXiv 2024

  13. [13]

    In: Text sum- marization branches out

    Lin, C.Y.: Rouge: A package for automatic evaluation of summaries. In: Text sum- marization branches out. pp. 74–81 (2004)

    work page 2004

  14. [14]

    Nature Biomedical Engineering1(9), 691–696 (2017)

    Maier-Hein, L., Vedula, S.S., Speidel, S., Navab, N., Kikinis, R., Park, A., Eisen- mann, M., Feussner, H., Forestier, G., Giannarou, S., et al.: Surgical data science for next-generation interventions. Nature Biomedical Engineering1(9), 691–696 (2017)

    work page 2017

  15. [15]

    In: European conference on computer vision

    Ni, B., Peng, H., Chen, M., Zhang, S., Meng, G., Fu, J., Xiang, S., Ling, H.: Expanding language-image pretrained models for general video recognition. In: European conference on computer vision. pp. 1–18. Springer (2022)

    work page 2022

  16. [16]

    Surgery today43(3), 271–275 (2013)

    Niitsu, H., Hirabayashi, N., Yoshimitsu, M., Mimura, T., Taomoto, J., Sugiyama, Y., Murakami, S., Saeki, S., Mukaida, H., Takiyama, W.: Using the objective struc- tured assessment of technical skills (osats) global rating scale to evaluate the skills of surgical trainees in the operating room. Surgery today43(3), 271–275 (2013)

    work page 2013

  17. [17]

    Medical Image Analysis78, 102433 (2022)

    Nwoye, C.I., Yu, T., Gonzalez, C., Seeliger, B., Mascagni, P., Mutter, D., Marescaux, J., Padoy, N.: Rendezvous: Attention mechanisms for the recognition of surgical action triplets in endoscopic videos. Medical Image Analysis78, 102433 (2022)

    work page 2022

  18. [18]

    Representation Learning with Contrastive Predictive Coding

    Oord, A.v.d., Li, Y., Vinyals, O.: Representation learning with contrastive predic- tive coding. arXiv preprint arXiv:1807.03748 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  19. [19]

    In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics

    Papineni, K., Roukos, S., Ward, T., Zhu, W.J.: Bleu: a method for automatic evaluation of machine translation. In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics. pp. 311–318 (2002)

    work page 2002

  20. [20]

    DINOv3

    Siméoni, O., Vo, H.V., Seitzer, M., Baldassarre, F., Oquab, M., Jose, C., Khali- dov, V., Szafraniec, M., Yi, S., Ramamonjisoa, M., et al.: Dinov3. arXiv preprint arXiv:2508.10104 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  21. [21]

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

    Song, E., Chai, W., Wang, G., Zhang, Y., Zhou, H., Wu, F., Chi, H., Guo, X., Ye, T., Zhang, Y., et al.: Moviechat: From dense token to sparse memory for long video understanding. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 18221–18232 (2024)

    work page 2024

  22. [22]

    Gemma 2: Improving Open Language Models at a Practical Size

    Team, G., Riviere, M., Pathak, S., Sessa, P.G., Hardin, C., Bhupatiraju, S., Hussenot, L., Mesnard, T., Shahriari, B., Ramé, A., et al.: Gemma 2: Improving open language models at a practical size. arXiv preprint arXiv:2408.00118 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  23. [23]

    IEEE transactions on medical imaging36(1), 86–97 (2016)

    Twinanda, A.P., Shehata, S., Mutter, D., Marescaux, J., De Mathelin, M., Padoy, N.: Endonet: a deep architecture for recognition tasks on laparoscopic videos. IEEE transactions on medical imaging36(1), 86–97 (2016)

    work page 2016

  24. [24]

    Advances in neural information pro- cessing systems30(2017)

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, Ł., Polosukhin, I.: Attention is all you need. Advances in neural information pro- cessing systems30(2017)

    work page 2017

  25. [25]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Vedantam, R., Lawrence Zitnick, C., Parikh, D.: Cider: Consensus-based image description evaluation. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 4566–4575 (2015)

    work page 2015

  26. [26]

    arXiv preprint arXiv:2501.11347 (2025)

    Wang, G., Bai, L., Wang, J., Yuan, K., Li, Z., Jiang, T., He, X., Wu, J., Chen, Z., Lei,Z.,etal.:Endochat:Groundedmultimodallargelanguagemodelforendoscopic surgery. arXiv preprint arXiv:2501.11347 (2025)

    work page arXiv 2025

  27. [27]

    IEEE Transactions on Circuits and Systems for Video Technology27(12), 2613–2622 (2016)

    Wang, P., Cao, Y., Shen, C., Liu, L., Shen, H.T.: Temporal pyramid pooling-based convolutional neural network for action recognition. IEEE Transactions on Circuits and Systems for Video Technology27(12), 2613–2622 (2016)

    work page 2016

  28. [28]

    Yang, A., Yang, B., Zhang, B., Hui, B., Zheng, B., Yu, B., Li, C., Liu, D., Huang, F., Wei, H., Lin, H., Yang, J., Tu, J., Zhang, J., Yang, J., Yang, J., Zhou, J., Lin, J., Dang, K., Lu, K., Bao, K., Yang, K., Yu, L., Li, M., Xue, M., Zhang, P., Zhu, Title Suppressed Due to Excessive Length 11 Q., Men, R., Lin, R., Li, T., Tang, T., Xia, T., Ren, X., Ren,...

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  29. [29]

    Advances in Neural Infor- mation Processing Systems37, 122952–122983 (2024)

    Yuan, K., Navab, N., Padoy, N., et al.: Procedure-aware surgical video-language pretraining with hierarchical knowledge augmentation. Advances in Neural Infor- mation Processing Systems37, 122952–122983 (2024)

    work page 2024

  30. [30]

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

    Yuan, K., Srivastav, V., Navab, N., Padoy, N.: Hecvl: Hierarchical video-language pretraining for zero-shot surgical phase recognition. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 306–316. Springer (2024)

    work page 2024

  31. [31]

    arXiv preprint arXiv:2512.09354 (2025)

    Zhao, X., Wang, Z., Zhang, Y., Cheng, G., Xu, Y., Deng, S., Liu, C., Wang, N., Yin, J.: Video-qtr: Query-driven temporal reasoning framework for lightweight video understanding. arXiv preprint arXiv:2512.09354 (2025)

    work page arXiv 2025