Pith. sign in

REVIEW 3 major objections 6 minor 35 references

A VLM with no payment training can classify bus payment types when the video is pre-ground to passenger clips and farebox close-ups.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-02 04:47 UTC pith:JFRAI3SV

load-bearing objection A readable proof-of-concept system with honest reporting of weak payment results, but its central claim that grounding improves VLM reasoning is untested — no baseline against ungrounded prompting. the 3 major comments →

arxiv 2607.13569 v2 pith:JFRAI3SV submitted 2026-07-15 cs.CV cs.AI

GHR-VLM: Making Zero-Shot Transit Video Analytics Realizable with Grounded Hybrid Reasoning

Kaicong Huang , Weiheng Oh , Jack M. Reilly , Thomas Guggisberg , Ruimin Ke This is my paper
classification cs.CV cs.AI
keywords transit video analyticsvision-language modelsvisual groundingedge-cloud collaborationzero-shot recognitionpassenger payment classificationbus surveillance video
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper claims that a general-purpose vision-language model, with no payment-specific training, can classify how a bus passenger pays—QR, cash, tap, swipe, or evasion—if it is shown compact, passenger-centered evidence instead of raw surveillance footage. The design splits the work across edge and cloud: lightweight edge modules detect door openings to isolate stops, track passengers, and crop passenger clips, while a cloud VLM first judges each passenger's facing direction and then classifies payment on a farebox-cropped contact sheet, refining with a second pass built from the frames the VLM itself selected as evidence. On real-world bus video, the pipeline reaches stop-detection F1 around 0.89, passenger-clip F1 around 0.70–0.85, and five-class payment accuracy around 0.31–0.54, with the worse numbers on the less consistently lit video. The paper's central point is that explicit spatiotemporal grounding is what makes zero-shot transit video analytics practical, and that the same grounding exposes how strongly fine-grained recognition still depends on video quality.

Core claim

GHR-VLM claims that invoking a VLM only on grounded passenger clips and farebox contact sheets makes zero-shot passenger-level payment classification realizable on real bus surveillance video. A door-grounded edge pipeline first filters the long stream down to stop intervals, then a rule-based tracker follows passengers and selects the largest tracked box as the boarding passenger nearest the farebox. A 'complex-to-simple' mapping lets the VLM judge only whether the person faces front, side, back, or inside, and a fixed rule converts those judgments to boarding, alighting, or staying. Retained clips are then submitted to two VLM passes over farebox-cropped contact sheets: the first samples s

What carries the argument

The load-bearing mechanism is spatiotemporal grounding, built from four pieces: (1) door-grounded stop filtering, which uses the front door's open/close state to cut the video into stop intervals; (2) rule-based passenger tracking with a projective prior—the largest tracked box is assumed to be the passenger nearest the farebox; (3) a complex-to-simple (CS) direction mapping that reduces boarding/alighting/staying to a generic facing-direction judgment (front/side→boarding, back→alighting, inside→staying); and (4) a two-stage coarse-to-fine payment classifier that crops a farebox region into a 16-frame contact sheet, asks the VLM to select its own evidence frames, then refines on a tighter c

Load-bearing premise

The whole chain rests on assuming the largest tracked box in the door area is the passenger who just boarded and is about to pay; if someone alighting lingers by the payment box, passengers board in a group, or the driver steps out, the payment classifier receives a wrong clip or no clip at all.

What would settle it

Run the pipeline on a stop where two passengers board together and where an alighting passenger remains beside the payment box; if passenger-clip F1 and payment accuracy do not fall compared with isolated single-boarder stops, then the largest-box and direction-assumption priors are not load-bearing. A simpler version: place a stationary dummy near the payment box while a passenger boards and check whether the clip sent to the classifier contains the wrong person.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Zero-shot transit payment analytics no longer requires payment-specific annotations, so the pipeline can be deployed to a new bus fleet, camera view, or agency without retraining.
  • Cloud inference is reduced to a few short VLM calls per stop, because only passenger clips and contact sheets leave the edge.
  • The two-stage evidence refinement can improve fine-grained recognition on well-lit footage—Swipe recall on the nighttime video rises from near zero to 0.333—showing that focused evidence helps a general VLM recover subtle actions.
  • The pipeline outputs structured stop, passenger, direction, and payment events, which transit agencies can use for crowding management, service planning, and revenue auditing.
  • The CS mapping means a general-purpose VLM does not need transit-specific action knowledge; it only needs to judge facing direction, a generic visual skill.

Where Pith is reading between the lines

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

  • The same grounding recipe—a fixed geometric prior, edge-side clip extraction, and an evidence-crop refinement—should transfer to other 'needle-in-a-haystack' fine-grained video tasks such as retail checkout verification or package handling, wherever a stable camera supplies a localization prior.
  • Because the pipeline assumes the largest tracked box is the boarding passenger, scenarios with group boardings, alighting passengers lingering by the farebox, or a driver leaving the cockpit will systematically corrupt the clip sent to the classifier; a stress test varying these scene factors would map the failure envelope.
  • At five-class accuracy of 0.31–0.54, full payment-method auditing is not yet production-ready; the binary evade/non-evade outputs (accuracy ~0.78–0.84) are closer to operationally usable and may be the right first deployment target.
  • Stage 2 refinement helped the well-lit video and hurt the poorly lit one, which suggests a cheap adaptive rule—keep the coarse prediction when visual quality or VLM confidence is low—could improve overall robustness.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. GHR-VLM proposes an edge-cloud framework for zero-shot payment classification on bus surveillance video. The edge pipeline detects stop intervals from door state, tracks passengers with SAM-based instance segmentation, and generates passenger clips; a VLM direction filter selects boarding passengers via a fixed orientation-to-behavior mapping; a two-stage farebox-cropped contact-sheet classifier assigns QR, cash, tap, swipe, or evade labels. Evaluation on two real videos reports stop F1 of 0.887/0.898, passenger-clip F1 of 0.702/0.847, five-class payment accuracy of 0.313/0.536, and binary evasion accuracy of 0.783/0.837. The paper claims that explicit visual grounding reduces cloud inference, avoids payment-specific training data, and provides the localized evidence that VLMs otherwise struggle to identify.

Significance. If the central claim were established, the framework would be practically valuable for transit agencies: it targets a real operational need, uses only real surveillance data, and proposes a plausible division of labor between lightweight edge modules and a cloud VLM. The paper is also honest about where the second stage hurts performance (C3_1) and about degraded video conditions. However, the core causal claim — that explicit grounding improves VLM reasoning — is not tested, because no comparison is made against an ungrounded VLM baseline or against a cost-equivalent full-input baseline. The empirical basis is also thin: two private videos, no confidence intervals, and no end-to-end passenger-level accuracy. The framework is promising, but the evidence does not yet support the 'realizable' language in the title and abstract.

major comments (3)
  1. [Sec. 4.3, Tables 1-2] The abstract and Sec. 1 claim that explicit visual grounding improves VLM reasoning and reduces cloud inference, but no experiment supports this. The only VLM comparison is Stage 1 vs Stage 2, both of which operate on farebox-cropped contact sheets built from the same grounded pipeline. There is no baseline in which the same VLM sees the full passenger clip, the full stop interval, or the raw video without grounding. Consequently, the observed accuracy differences are confounded with input-length reduction and cannot establish that the grounded spatiotemporal evidence itself helps. The paper should include such a baseline and also quantify token/cost savings, since 'reduces cloud inference' is an asserted benefit that is never measured.
  2. [Sec. 3.2-3.3, Eqs. (5) and (9)] The pipeline assumes that the largest tracked box is the boarding passenger nearest the farebox (Eq. 5) and that a VLM orientation judgment maps deterministically to boarding/alighting/staying (Eq. 9: front/side → boarding, back → alighting, inside → staying). These assumptions are fragile in exactly the situations the paper itself acknowledges: the driver leaving the cockpit, simultaneous boarding events, and passengers reappearing (Sec. 4.2). When Eq. (5) selects the wrong identity or Eq. (9) filters out a true boarder, no passenger clip reaches the payment classifier, and the error propagates to the final output. The paper reports component-level stop and clip F1, but not the joint passenger-level payment accuracy. The authors should report the end-to-end success rate and analyze these failure modes quantitatively.
  3. [Sec. 4.1, 4.3, Tables 1-2] The empirical evidence is limited to two private videos with point estimates and no confidence intervals. The abstract states 486 minutes, while Sec. 4.1 says 436 minutes, a discrepancy that should be resolved. On C3_1, Stage 2 reduces five-class accuracy from 0.349 to 0.313 and binary evasion accuracy from 0.813 to 0.783; the paper attributes this to visual quality, but no statistical analysis is provided. Given the title's promise of making zero-shot transit video analytics 'realizable', the authors should either provide more data, per-stop or per-error breakdowns, or at least temper the claim to match the two-video, unstable results.
minor comments (6)
  1. [Sec. 4.1] The duration inconsistency (486 min vs 436 min vs '9 hours') should be corrected in the final version.
  2. [Sec. 3.2] Please define 'alternate frames' explicitly (e.g., every 2nd frame at 10 FPS) and clarify the relationship between sampled-frame indices and absolute timestamps in Eq. (6)-(7).
  3. [Sec. 3.4] The margin notation '(0.35,0.35,-0.50)' is unclear: are these absolute normalized offsets or fractions of the box dimensions? Please define the coordinate convention once and reuse it.
  4. [Sec. 3.4] When Stage 1 returns a non-contiguous evidence-frame set, Stage 2 uses the earliest and latest frames, which may include unrelated frames. Please describe how evidence frames are combined or justify this choice.
  5. [Sec. 4] No prompt templates, model version details, or code are provided. Given the emphasis on zero-shot behavior, releasing the exact prompts would be important for reproducibility.
  6. [Sec. 4.3, Table 2] Class-wise counts are small and arrows indicate only relative changes; please provide per-class sample sizes or confidence intervals to help judge the stability of the reported improvements.

Circularity Check

0 steps flagged

No significant circularity: the paper is an empirical systems evaluation, and its self-citations are not load-bearing premises.

full rationale

GHR-VLM is an empirical systems paper rather than a derivation chain. The claimed outputs—stop intervals, passenger clips, and payment labels—are obtained by running explicit pipelines (door-grounded stop filtering, rule-based tracking, VLM direction mapping, two-stage farebox-grounded payment classification) and are evaluated against manually annotated ground truth on two videos. No equation reduces a target output to a fitted parameter: thresholds such as θ_I=0.20, the 0.5 s clip minimum, the 15 s stop merge, and the stage-specific margins are hand-set engineering choices, and the reported numbers are direct measurements, not fitted values renamed as predictions. The self-citations [10] and [11] are used only as related-work references (e.g., motivation for fine-grained transit data and examples of supervised video models); they are not invoked as proof of the architecture's effectiveness, and no uniqueness theorem or prior ansatz is imported from them. The paper's central empirical weakness—the absence of a baseline in which the same VLM sees uncropped or ungrounded video—undermines the causal claim that explicit grounding improves VLM reasoning, but a missing control is not circularity. Similarly, the abstract's 486 minutes versus Section 4.1's 436 minutes is a reporting inconsistency, not a circular derivation. The admitted limitations (driver leaving cockpit, simultaneous boarding, degraded illumination) are acknowledged failure modes rather than hidden restatements of the method's assumptions. Overall, no step in the paper's chain is equivalent by construction to its inputs.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 0 invented entities

The paper introduces no new physical or formal entities; its artifacts are representations (contact sheets, clips) and hand-set thresholds. The main external reliance is on proprietary models whose prompts are undisclosed.

free parameters (4)
  • Temporal IoU match threshold θ_I
    §3.2 Eq. 4: cross-batch identity match threshold 0.20; hand-chosen; controls identity inheritance and hence clip boundaries.
  • Clip minimum duration and merge window
    §3.2: clips <0.5 s removed; adjacent run boundaries within 3 sampled frames merged; hand-chosen; directly sets passenger-clip recall/precision.
  • Stop-merge window
    §4.2 Round 2 merges stops whose retained clips are separated by <15 s; hand-chosen on the same two videos; improves stop F1 at recall cost.
  • Farebox crop margins
    §3.4: Stage1 margins (0.35,0.35,−0.50), Stage2 (0,0.10,−0.50); hand-chosen to control the evidence presented to the payment VLM.
axioms (4)
  • domain assumption Only boarding passengers interact with the farebox and the camera is fixed and inward-facing, so the direction mapping ψ_CS (Eq. 9) is valid.
    §3.3, Eq. 9: front/side→boarding, back→alighting, inside→staying; if an alighting passenger pauses near the farebox or a boarding passenger is seen from behind, clips are mislabeled or dropped.
  • domain assumption The largest tracked box is the passenger nearest the farebox / main passenger.
    §3.2 Eq. 5: m_i = argmax area; the text calls this 'closest to the fare area' but uses area; group boarding and occlusion break this, as the paper's own error analysis admits.
  • domain assumption The farebox location detected on the first passenger clip remains fixed for the whole video.
    §3.4: B_f is cached from P_{k0,j0}[0]; camera shake or reframing would make the payment crops miss the farebox.
  • domain assumption External models (SAM3, GPT-4o, GPT-5.4-mini) provide reliable grounded localization and zero-shot classification as invoked.
    §4.1: The entire pipeline depends on these closed, unshipped models; prompts are not included, so this cannot be independently checked.

pith-pipeline@v1.3.0-alltime-deepseek · 9822 in / 13500 out tokens · 131025 ms · 2026-08-02T04:47:01.782561+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of GHR-VLM: Making Zero-Shot Transit Video Analytics Realizable with Grounded Hybrid Reasoning." pith.science (2026). https://pith.science/paper/JFRAI3SV

@misc{pith2026260713569,
  author       = {Pith},
  title        = {Pith review of: GHR-VLM: Making Zero-Shot Transit Video Analytics Realizable with Grounded Hybrid Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JFRAI3SV}},
  note         = {Machine review of arXiv:2607.13569}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Transit video understanding can provide valuable fine-grained data that conventional passenger counters and fare systems cannot capture. However, supervised video models require task-specific annotations, while applying vision-language models (VLMs) directly to long onboard videos is unreliable and costly. To leverage the complementary strengths of both approaches, we propose GHR-VLM, a visual grounded hybrid reasoning framework for zero-shot transit-bus video analytics. It is motivated by the observation that explicit visual grounding can improve VLM reasoning by converting long surveillance streams into compact, passenger-centered spatiotemporal evidence. Specifically, we propose an edge-cloud design in which a lightweight edge-based monitor continuously tracks door status and segments passenger clips. A backend VLM then identifies boarding passengers and classifies payment behavior through a two-stage coarse-to-fine refinement of spatiotemporal evidence. By invoking the VLM only on grounded passenger clips and contact sheets, GHR-VLM reduces cloud inference, avoids payment-specific training data, and supplies the localized evidence that VLMs otherwise struggle to identify. Evaluation on 486 minutes of real-world bus surveillance video demonstrates the potential of grounded edge-cloud reasoning for passenger-level payment analytics while highlighting the challenges posed by degraded video conditions.

Figures

Figures reproduced from arXiv: 2607.13569 by Jack M. Reilly, Kaicong Huang, Ruimin Ke, Thomas Guggisberg, Weiheng Oh.

Figure 1
Figure 1. Figure 1: Comparison of model-based, VLM-based, and our grounded hybrid approaches for transit video analytics. Model [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of GHR-VLM. The edge pipeline grounds the front-door region and restricts downstream processing to stop [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Full-video timelines of detected stops and passenger clips for C3_1 (top) and C3_3 (bottom). Each panel compares the [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Payment-type confusion matrices for both VLM stages on C3_1 and C3_3. Each cell reports the number of samples [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

35 extracted references · 8 linked inside Pith

  1. [1]

    Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, and Cordelia Schmid. 2021. Vivit: A video vision transformer. InProceedings of the IEEE/CVF international conference on computer vision. 6836–6846

    2021

  2. [2]

    Jinze Bai, Shuai Bai, Shusheng Yang, Shijie Wang, Sinan Tan, Peng Wang, Jun- yang Lin, Chang Zhou, and J Qwen-VL Zhou. 2023. A versatile vision-language model for understanding, localization, text reading, and beyond.arXiv preprint arXiv:2308.129666 (2023), 3

    Pith/arXiv arXiv 2023

  3. [3]

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, Lianghao Deng, Wei Ding, Chang Gao, Chunjiang Ge, et al. 2025. Qwen3-vl technical report.arXiv preprint arXiv:2511.21631(2025)

    Pith/arXiv arXiv 2025

  4. [4]

    Gedas Bertasius, Heng Wang, and Lorenzo Torresani. 2021. Is space-time atten- tion all you need for video understanding?. InIcml, Vol. 2. 4

    2021

  5. [5]

    Nicolas Carion, Laura Gustafson, Yuan-Ting Hu, Shoubhik Debnath, Ronghang Hu, Didac Suris, Chaitanya Ryali, Kalyan Vasudev Alwala, Haitham Khedr, An- drew Huang, et al. 2025. Sam 3: Segment anything with concepts.arXiv preprint arXiv:2511.16719(2025)

    Pith/arXiv arXiv 2025

  6. [6]

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. 2020. End-to-end object detection with trans- formers. InEuropean conference on computer vision. Springer, 213–229

    2020

  7. [7]

    José Correa, Tobias Harks, Vincent JC Kreuzen, and Jannik Matuschke. 2017. Fare evasion in transit networks.Operations research65, 1 (2017), 165–183

    2017

  8. [8]

    Amir Dib, Noëlie Cherrier, Martin Graive, Baptiste Rérolle, and Eglantine Schmitt. 2023. Unified occupancy on a public transport network through com- bination of AFC and APC data. In2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC). IEEE, 1963–1970

    2023

  9. [9]

    Christoph Feichtenhofer, Haoqi Fan, Jitendra Malik, and Kaiming He. 2019. Slow- fast networks for video recognition. InProceedings of the IEEE/CVF international conference on computer vision. 6202–6211

    2019

  10. [10]

    Kaicong Huang, Talha Azfar, Jack Reilly, and Ruimin Ke. 2025. Transitreid: Transit od data collection with occlusion-resistant dynamic passenger re- identification.arXiv preprint arXiv:2504.11500(2025)

    Pith/arXiv arXiv 2025

  11. [11]

    Kaicong Huang, Weiheng Oh, Thomas Guggisberg, and Ruimin Ke. 2026. iPay: Integrated Payment Action Recognition via Multimodal Networks and Adaptive Spatial Prior Learning.arXiv preprint arXiv:2605.10732(2026)

    Pith/arXiv arXiv 2026

  12. [12]

    Nico Jahn and Michael Siebert. 2022. Engineering the neural automatic passen- ger counter.Engineering Applications of Artificial Intelligence114 (2022), 105148

    2022

  13. [13]

    Glenn Jocher, Jing Qiu, and Ayush Chaurasia. 2024. Ultralytics yolo11

    2024

  14. [14]

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al

  15. [15]

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023. Blip-2: Bootstrap- ping language-image pre-training with frozen image encoders and large lan- guage models. InInternational conference on machine learning. PMLR, 19730– 19742

    2023

  16. [16]

    Liunian Harold Li, Pengchuan Zhang, Haotian Zhang, Jianwei Yang, Chunyuan Li, Yiwu Zhong, Lijuan Wang, Lu Yuan, Lei Zhang, Jenq-Neng Hwang, et al

  17. [17]

    Bin Lin, Yang Ye, Bin Zhu, Jiaxi Cui, Munan Ning, Peng Jin, and Li Yuan. 2024. Video-llava: Learning united visual representation by alignment before projec- tion. InProceedings of the 2024 conference on empirical methods in natural lan- guage processing. 5971–5984

    2024

  18. [18]

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023. Visual in- struction tuning.Advances in neural information processing systems36 (2023), 34892–34916

    2023

  19. [19]

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Qing Jiang, Chunyuan Li, Jianwei Yang, Hang Su, et al. 2024. Grounding dino: Marry- ing dino with grounded pre-training for open-set object detection. InEuropean conference on computer vision. Springer, 38–55

    2024

  20. [20]

    Muhammad Maaz, Hanoona Rasheed, Salman Khan, and Fahad Khan. 2024. Video-chatgpt: Towards detailed video understanding via large vision and lan- guage models. InProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 12585–12602

    2024

  21. [21]

    Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, et al. 2022. Simple open-vocabulary object detection. InEuropean conference on computer vision. Springer, 728–755

    2022

  22. [22]

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman Rädle, Chloe Rolland, Laura Gustafson, et al

  23. [23]

    Tianhe Ren, Shilong Liu, Ailing Zeng, Jing Lin, Kunchang Li, He Cao, Jiayu Chen, Xinyu Huang, Yukang Chen, Feng Yan, et al. 2024. Grounded sam: Assem- bling open-world models for diverse visual tasks.arXiv preprint arXiv:2401.14159 (2024)

    Pith/arXiv arXiv 2024

  24. [24]

    Wei-Qing Ren, Yu-Ben Qu, Chao Dong, Yu-Qian Jing, Hao Sun, Qi-Hui Wu, and Song Guo. 2023. A survey on collaborative DNN inference for edge intelligence. Machine Intelligence Research20, 3 (2023), 370–395

    2023

  25. [25]

    Enxin Song, Wenhao Chai, Guanhong Wang, Yucheng Zhang, Haoyang Zhou, Feiyang Wu, Haozhe Chi, Xun Guo, Tian Ye, Yanting Zhang, et al. 2024. Moviechat: From dense token to sparse memory for long video understanding. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition. 18221–18232

    2024

  26. [26]

    Zhan Tong, Yibing Song, Jue Wang, and Limin Wang. 2022. Videomae: Masked autoencoders are data-efficient learners for self-supervised video pre-training. Advances in neural information processing systems35 (2022), 10078–10093

    2022

  27. [27]

    Johannes van der Vyver. 2024. A Deep Neural Network Approach to Fare Eva- sion.arXiv preprint arXiv:2405.17855(2024)

    Pith/arXiv arXiv 2024

  28. [28]

    Mengmeng Wang, Jiazheng Xing, and Yong Liu. 2021. Actionclip: A new para- digm for video action recognition.arXiv preprint arXiv:2109.08472(2021)

    Pith/arXiv arXiv 2021

  29. [29]

    Nicolai Wojke, Alex Bewley, and Dietrich Paulus. 2017. Simple online and real- time tracking with a deep association metric. In2017 IEEE international confer- ence on image processing (ICIP). IEEE, 3645–3649

    2017

  30. [30]

    Sijie Yan, Yuanjun Xiong, and Dahua Lin. 2018. Spatial temporal graph convo- lutional networks for skeleton-based action recognition. InProceedings of the AAAI conference on artificial intelligence, Vol. 32

    2018

  31. [31]

    Hang Zhang, Xin Li, and Lidong Bing. 2023. Video-llama: An instruction-tuned audio-visual language model for video understanding. InProceedings of the 2023 conference on empirical methods in natural language processing: system demon- strations. 543–553

    2023

  32. [32]

    Yifu Zhang, Peize Sun, Yi Jiang, Dongdong Yu, Fucheng Weng, Zehuan Yuan, Ping Luo, Wenyu Liu, and Xinggang Wang. 2022. Bytetrack: Multi-object track- ing by associating every detection box. InEuropean conference on computer vi- sion. Springer, 1–21

    2022

  33. [2022]

    InProceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Grounded language-image pre-training. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 10965–10975

  34. [2023]

    InProceedings of the IEEE/CVF international conference on computer vision

    Segment anything. InProceedings of the IEEE/CVF international conference on computer vision. 4015–4026

  35. [2025]

    InInternational Conference on Learning Representations, Vol

    Sam 2: Segment anything in images and videos. InInternational Conference on Learning Representations, Vol. 2025. 28085–28128

    2025