Pith. sign in

REVIEW 4 major objections 4 minor 44 references

Enhanced Multi-Object Tracking Using Pose-based Virtual Markers in 3x3 Basketball

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Overlaying pose-based virtual markers with known player identities onto 3x3 basketball video lets a single pose model track all players through heavy occlusion with zero identity switches, beating automated trackers by more than ten HOTA…

desk verdict The headline HOTA gain is an artifact of test-time identity markers; the active-learning result is real but secondary. read the letter →

arxiv 2412.06258 v1 pith:BNVFTMGB submitted 2024-12-09 cs.CV

Li Yin , Calvin Yeung , Qingrui Hu , Jun Ichikawa , Hirotsugu Azechi , Susumu Takahashi , Keisuke Fujii This is my paper · ORCID
classification cs.CV
keywords multi-objecttrackingposeestimationvirtualmarkers3x3basketballactivelearningocclusionhandlingHOTAidentityswitches
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

This paper tries to establish that in team-sports tracking, the hard problem of keeping players' identities straight under occlusion and near-identical uniforms can be bypassed by painting virtual markers—colored keypoints with known IDs—onto the video, and then tracking those markers. On the authors' 3x3 basketball dataset, the method reports a HOTA score (a tracking metric balancing detection and association accuracy) of 72.3%, more than ten points above recent fully automated trackers, with zero identity switches; the results table lists 72.6 ± 2.7. The paper also shows that active learning on only 659 of 3,817 frames yields pose accuracy nearly matching full annotation (4.40 vs 4.16 pixels test error), cutting labeling cost. The central claim is that pose-based virtual markers, not learned appearance association, are the effective mechanism for robust multi-object tracking in crowded, similar-appearance scenes.

What carries the argument

The central object is the virtual marker (VM): a colored dot or set of dots overlaid on video frames at known player keypoints, carrying identity through color. The pipeline has three steps: a multi-agent pose estimator is trained with active learning to generate VMs; those VMs are overlaid on training videos to force ID-consistent annotation and on test videos to give the tracker identity cues; then a single-agent pose estimator outputs six keypoints per player, which are converted to bounding boxes with a padding offset. The VMs do the association work that appearance-based re-identification cannot: they make 'who is who' visible in the image, leaving the tracker with the simpler task of precise keypoint localization rather than identity reasoning.

What would settle it

Run the same pose-tracking and bounding-box pipeline on the same 3x3 basketball test videos with all virtual-marker overlays removed; if HOTA falls to the level of the automated baselines, the reported gain and zero ID switches come from the injected identity cues rather than from the tracking architecture.

Watch

Extended reading notes

Core claim

The central claim is that a pose-based virtual-marker pipeline can solve multi-object tracking in team sports under heavy occlusion and appearance similarity. The method trains a multi-agent pose estimator to generate colored keypoint markers with known player identities, overlays those markers on both training and test videos, and then tracks each player's six keypoints with a single-agent pose estimator; the keypoints are converted into bounding boxes for comparison with automated MOT methods. On the authors' 3x3 basketball dataset, this pipeline achieves a HOTA of 72.3% (72.6 ± 2.7 in Table 2), more than ten points above marker-free baselines, with zero identity switches. The authors attribute the gain to the markers providing a direct visual identity cue, which keeps association robust when players occlude each other or wear identical uniforms, while active learning reduces the annotation burden.

Load-bearing premise

The load-bearing premise is that overlaying virtual markers with known player IDs onto the test videos is a legitimate part of the tracking pipeline, so the comparison with fully automated trackers that do not receive these identity cues measures the same task.

Editorial extensions

If this is right

  • If the reported results hold, adding pose-based virtual markers to team-sports video raises HOTA by more than ten points over fully automated trackers and eliminates identity switches in 3x3 basketball.
  • Active learning on occlusion-heavy frames cuts the annotated training set by roughly a factor of six (659 vs 3,817 frames) while keeping test error within 0.24 pixels of full annotation.
  • Smaller markers work better: 1-pixel VMs outperform 3-pixel VMs at every marker count, and six VMs per player outperform one or three, so dense, unobtrusive identity cues are the recommended configuration.
  • The way keypoints are converted to bounding boxes matters as much as the tracking itself: the padding method scores HOTA 75.0 versus 43.8 for the max-min method, so the conversion step is a major component of the reported gains.

Reading between the lines

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

  • If overlaying identity cues at test time is accepted, the method reframes sports MOT as a marker-tracking problem: the hard part shifts from data association to precise keypoint localization, so improvements in pose stability translate directly into tracking scores.
  • The paper leaves open whether the same advantage would appear on public benchmarks, which do not provide pre-assigned player identities to overlay; a test on a public dataset with markers generated from its own pose annotations would indicate generalizability.
  • The zero-ID-switch result is partly a consequence of the markers being visible in the input: if the pose model can read the marker color, identity is supplied by the image rather than inferred, so the metric may be measuring marker-visibility fidelity as much as tracking ability.
  • Because active learning selected occlusion-heavy frames for annotation, the approach could adapt to other fixed-camera team sports such as futsal or handball, where players are few, uniforms are similar, and occlusions cluster around the ball.
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

4 major / 4 minor

Summary. The paper proposes Sports-vmTracking, an adaptation of the vmTracking multi-animal pose-tracking method to 3x3 basketball. The pipeline has three steps: (1) train a multi-agent DeepLabCut model to detect six keypoints per player and use these to overlay colored virtual markers (VMs) onto video frames; (2) train a single-agent pose model on the VM-augmented videos; (3) run that model on VM-augmented test videos, convert the predicted keypoints to bounding boxes, and evaluate standard MOT metrics (HOTA, DetA, AssA, FN, FP, ID switches). The authors report a HOTA of 72.6 (abstract says 72.3), about 10 points above Deep-EIoU, BOT-SORT, Basketball-SORT, and maDLC without VMs, and 0 ID switches. They also report an active-learning result in which 659 annotated frames achieve a test RMSE of 4.40 px versus 4.16 px with all 3,817 frames annotated. The central empirical claim is that the VM-based method substantially outperforms automated MOT trackers on a private 3x3 basketball dataset.

Significance. If the reported comparison were valid, the paper would demonstrate a practically important approach to occlusion- and appearance-challenged sports tracking, and the active-learning result would be a useful efficiency contribution for pose annotation. The authors are explicit about the pipeline and disclose that keypoint-derived bounding boxes are 'not accurate' in Section 3.3. However, the significance is undercut by a fundamental evaluation confound: Sports-vmTracking is tested on videos with identity-carrying virtual markers overlaid, while the compared automated methods are tested on raw marker-free video. The 0 ID switches and the large HOTA gap therefore reflect the input differences at least as much as any tracking capability. The active-learning comparison is a separate, plausible contribution, but it does not validate the claimed MOT superiority. The paper does not release code or data, which limits reproducibility.

major comments (4)
  1. [Section 3.3, Section 4.2, Table 2] The headline MOT comparison is confounded. Sports-vmTracking is evaluated on test videos onto which virtual markers with known player identities have been overlaid (Section 4.2: 'In the test video dataset with VM, we created 6 test datasets with varying sizes and quantities of VM'), and Step 3 in Section 3.3 performs inference on these VM-labeled videos. The single-agent pose model is trained on VM videos where ID-consistent labeling is 'guided by the VMs' (Section 3.2). The baselines (Deep-EIoU, BOT-SORT, Basketball-SORT, maDLC w/o VMs) are evaluated on raw marker-free video. The test inputs therefore differ in a way that directly supplies identity cues to the proposed method, so the reported 0.0 ID switches, the near-zero FN/FP, and the >10-point HOTA advantage in Table 2 do not establish that Sports-vmTracking is better at the MOT task faced by the baselines.
  2. [Abstract and Table 2] The abstract reports an average HOTA of 72.3%, but Table 2 reports 72.6 ± 2.7 for Sports-vmTracking, and Table 3 reports 72.6 for the 1-pixel, 6-point configuration. The 72.3 value appears nowhere in the tables. The authors should state exactly which configuration and which number is the headline result, and reconcile the discrepancy.
  3. [Section 3.3 and Table 4] The bounding-box conversion step is itself a source of major uncertainty. Section 3.3 states that the converted keypoint boxes 'are not accurate,' and Table 4 shows that when ground-truth keypoints are used, the padding method reaches HOTA 75.0, whereas the Max_Min method reaches only 43.8. This indicates that the final MOT quality is highly sensitive to the keypoint-to-box conversion and to keypoint accuracy, not merely to association. However, all reported Sports-vmTracking HOTA numbers (72.6 in Table 2) are below the ground-truth-keypoint upper bound, and the paper does not quantify how much of the 72.6 is attributable to the pose model versus the conversion method. The comparison would be much more informative if the same bounding-box conversion were applied to the keypoint outputs of the baselines, or if a marker-free variant of the proposed pipeline were reported.
  4. [Section 4.2, Table 1] The active-learning result is presented as a main contribution, but it does not validate the MOT claims. Table 1 shows that 659 annotated frames yield a test RMSE of 4.40 px versus 4.16 px with 3,817 frames, which is a reasonable efficiency outcome. However, this pose model is trained and tested on VM-augmented data, and the MOT evaluation is on the same VM-augmented test set. The active-learning result would be more convincing if the pose error were reported for marker-free test frames as well, or if the downstream MOT metrics were reported as a function of annotation budget.
minor comments (4)
  1. [Section 4.3] The text uses 'Sport-vmTracking' in one paragraph ('Sport-vmTracking’s HOTA score exceeds...') while the rest of the paper uses 'Sports-vmTracking'; please unify the name.
  2. [Table 1 caption] The table caption reads 'T able 1' with an extra space; please fix the formatting.
  3. [Section 4.4] The sentence 'In vmTracking, the VM size was set to a minimum of 1 pixel' is unclear: it presumably means a single-pixel marker, but 'minimum' could be misread as a lower bound; please rephrase.
  4. [Section 3.1] The description of the VM creation step says 'After manually correcting some ID switches...', which is a manual intervention step, but the paper later claims an 'automated' pipeline; the role of manual correction in the test-time procedure should be clarified.

Circularity Check

1 steps flagged · score 8.0 of 10

The central benchmark is confounded: Sports-vmTracking's test videos carry overlaid colored markers with known player IDs, so the reported 0 ID switches and >10-point HOTA advantage are inherited from the test input rather than produced by multi-object tracking.

  1. self definitional [Sec. 3.1 (Creation of Virtual Markers); Sec. 3.3 (Step 3); Sec. 4.3 (Benchmark Results)]
    "After manually correcting some ID switches and ensuring the correct colors are assigned to distinguish individuals, all 6 or a subset of the human keypoints will be selected as VMs... These VMs are then overlaid onto the raw video to generate a VM video. ... The single-agent pose model trained in Step 2 is deployed in the VM-labeled test videos to generate keypoints for each player, identified by a unique ID."

    The VMs are colored markers placed at selected keypoints of known players; Section 3.1 states that colors are assigned to distinguish individuals, and Step 2 trains on 'VM videos, with ID-consistent labeling between frames guided by the VMs.' The Step 3 model therefore never performs data association: each player's identity is supplied by the marker color and by per-player output channels (player1_head, etc.). The reported '0 ID switches' and near-perfect FN/FP are forced by this input, and the comparison baselines (Deep-EIoU, BOT-SORT, Basketball-SORT, maDLC) are explicitly evaluated on videos 'without the use of VMs.' This is a comparison of two different tasks, so the >10-point HOTA gap and 0 IDs reduce to the test-input construction, not to MOT performance.

full rationale

The central MOT claim is not self-contained because the test input already contains the identities that the method then reports. Section 3.1 describes selecting VMs after 'manually correcting some ID switches and ensuring the correct colors are assigned to distinguish individuals' and overlaying them on raw video; Section 3.3 then uses those VM-labeled videos as the test set for the single-agent pose model, whose outputs are keypoints 'identified by a unique ID.' The ID is therefore supplied by the marker color and by the per-player output channels, not by any association algorithm. The baseline trackers are run on the same videos without VMs, so the headline comparison is between different tasks; the 0 ID switches and >10-point HOTA advantage are an artifact of the marker injection. The paper itself concedes in Section 3.3 that the resulting bounding boxes 'are not accurate' and in Section 5 notes the omission of hand keypoints; these limitations are acknowledged but do not repair the identity confound. The active-learning efficiency comparison (659 vs 3817 frames, 4.40 vs 4.16 px RMSE) is independent and non-circular, but it does not establish the claimed MOT superiority.

Assumptions & free parameters 5 free parameters · 4 assumptions · 1 invented entities

The central claim rests on the assumption that injecting identity-carrying markers into test videos is a valid MOT intervention. The method adds several hand-chosen parameters (VM size, VM quantity, padding, threshold) that are tuned on the test set, and the dataset is private and drawn from co-authored work.

free parameters (5)
  • VM size (pixels) = 1 (best in Table 3)
    Chosen from a test-set sweep of 1 vs 3 pixels; affects tracking performance.
  • VM quantity per player = 6 (best in Table 3)
    Chosen from a test-set sweep of 1, 3, or 6 points; 6 points at 1 px gave the best HOTA.
  • Bounding box padding offset = not specified
    Used to convert keypoints to bboxes (Section 4.5); the offset size is not reported, so replication is impossible.
  • ID-switch exclusion threshold = not specified
    Euclidean distance threshold used to discard keypoints with detected ID switches (Section 3.3); value not reported.
  • Active learning annotation frame counts = 691 (multi-agent), 659 (single-agent)
    These frame counts are a design choice; the paper shows 659 frames achieve test error 4.40 vs 4.16 for 3,817 frames.
assumptions (4)
  • domain assumption DLCRNet and EfficientNet_b0 provide accurate keypoint predictions.
    The entire pipeline assumes the pose models detect keypoints and VMs reliably (Sections 3.1-3.2).
  • ad hoc to paper Virtual markers overlaid on test videos are visible and detectable by the trained single-agent model.
    The method's tracking performance depends on markers appearing in the test video (Section 4.2).
  • domain assumption Keypoints can be converted into bounding boxes with sufficient accuracy for HOTA evaluation.
    Used in Section 3.3 and 4.5; the paper itself notes inaccuracy.
  • domain assumption Ground-truth bounding boxes used for HOTA are correctly annotated.
    HOTA evaluation on the private dataset assumes correct GT boxes (Section 4.1).
invented entities (1)
  • Virtual markers (VMs) independent evidence
    purpose: Colored dots overlaid on players to provide identity cues for tracking.
    The markers are visible in the video and the model detects them; they are a new construct introduced to the sports domain, building on vmTracking.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enhanced Multi-Object Tracking Using Pose-based Virtual Markers in 3x3 Basketball." pith.science (2026). https://pith.science/paper/BNVFTMGB

@misc{pith2026241206258,
  author       = {Pith},
  title        = {Pith review of: Enhanced Multi-Object Tracking Using Pose-based Virtual Markers in 3x3 Basketball},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BNVFTMGB}},
  note         = {Machine review of arXiv:2412.06258}
}
read the original abstract

Multi-object tracking (MOT) is crucial for various multi-agent analyses such as evaluating team sports tactics and player movements and performance. While pedestrian tracking has advanced with Tracking-by-Detection MOT, team sports like basketball pose unique challenges. These challenges include players' unpredictable movements, frequent close interactions, and visual similarities that complicate pose labeling and lead to significant occlusions, frequent ID switches, and high manual annotation costs. To address these challenges, we propose a novel pose-based virtual marker (VM) MOT method for team sports, named Sports-vmTracking. This method builds on the vmTracking approach developed for multi-animal tracking with active learning. First, we constructed a 3x3 basketball pose dataset for VMs and applied active learning to enhance model performance in generating VMs. Then, we overlaid the VMs on video to identify players, extract their poses with unique IDs, and convert these into bounding boxes for comparison with automated MOT methods. Using our 3x3 basketball dataset, we demonstrated that our VM configuration has been highly effective, and reduced the need for manual corrections and labeling during pose model training while maintaining high accuracy. Our approach achieved an average HOTA score of 72.3%, over 10 points higher than other state-of-the-art methods without VM, and resulted in 0 ID switches. Beyond improving performance in handling occlusions and minimizing ID switches, our framework could substantially increase the time and cost efficiency compared to traditional manual annotation.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 30 canonical work pages

  1. [1]

    In: European Conference on Computer Vision, pp

    Zhang, Y., Sun, P., Jiang, Y., Yu, D., Weng, F., Yuan, Z., Luo, P., Liu, W., Wang, X.: Bytetrack: Multi-object tracking by associating every detection box. In: European Conference on Computer Vision, pp. 1–21 (2022). Springer

    2022

  2. [2]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Meinhardt, T., Kirillov, A., Leal-Taixé, L., Feichtenhofer, C.: Trackformer: Multi- object tracking with transformers. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 8844–8854 (2022)

    work page 2022

  3. [3]

    arXiv preprint arXiv:2012.15460 (2020)

    Sun, P., Cao, J., Jiang, Y., Cheng, Z., Zhang, B., Xie, D., Yuan, Z.: Transtrack: Multiple-object tracking with transformer. arXiv preprint arXiv:2012.15460 (2020)

    arXiv 2020

  4. [4]

    arXiv preprint arXiv:2206.14651 (2022) 14

    Aharon, N., Ben-Ari, R.: Bot-sort: Robust associations multi-pedestrian tracking. arXiv preprint arXiv:2206.14651 (2022) 14

    arXiv 2022

  5. [5]

    In: Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp

    Ren, B.-H., Wang, S.-W., Wang, M.-H., Lee, W.-J.: Tracknet: A deep learning network for tracking high-speed and tiny objects in sports applications. In: Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp. 3–10 (2018)

    work page 2018

  6. [6]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Liu, X., Wu, F., Wang, S.: Stam: A spatio-temporal attention mechanism for multi-object tracking in sports. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1238–1247 (2021)

    work page 2021

  7. [7]

    Vats, K., Walters, P., Fani, M., Clausi, D.A., Zelek, J.S.: Player tracking and identificationinicehockey.ExpertSystemswithApplications 213,119250(2023)

    work page 2023

  8. [8]

    Multimedia Tools and Applications, 1–17 (2024)

    Hu, Q., Scott, A., Yeung, C., Fujii, K.: Basketball-sort: an association method for complex multi-object occlusion problems in basketball multi-object tracking. Multimedia Tools and Applications, 1–17 (2024)

    work page 2024

Show all 44 references
  1. [9]

    In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp

    Huang, H.-W., Yang, C.-Y., Sun, J., Kim, P.-K., Kim, K.-J., Lee, K., Huang, C.- I., Hwang, J.-N.: Iterative scale-up expansioniou and deep features association for multi-object tracking in sports. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vi...

    2024

  2. [10]

    arXiv preprint arXiv:2304.05170 (2023)

    Cui,Y.,Zeng,C.,Zhao,X.,Yang,Y.,Wu,G.,Wang,L.:Sportsmot:Alargemulti- object tracking dataset in multiple sports scenes. arXiv preprint arXiv:2304.05170 (2023)

    2023 arXiv

  3. [11]

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

    Cioppa, A., Giancola, S., Deliege, A., Kang, L., Zhou, X., Cheng, Z., Ghanem, B., Van Droogenbroeck, M.: Soccernet-tracking: Multiple object tracking dataset and benchmark in soccer videos. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, ...

    2022

  4. [12]

    arXiv preprint arXiv:submit/5550700 (2023)

    Scott, A., Uchida, I., Ding, N., Umemoto, R., Bunker, R., Kobayashi, R., Koyama, T., Onishi, M., Kameda, Y., Fujii, K.: Teamtrack: An algorithm and benchmark dataset for multi-sport multi-object tracking in full-pitch videos. arXiv preprint arXiv:submit/5550700 (2023)

    2023

  5. [13]

    Naval Research Logistics Quarterly 2(1-2), 83–97 (1955)

    Kuhn, H.W.: The hungarian method for the assignment problem. Naval Research Logistics Quarterly 2(1-2), 83–97 (1955)

    1955

  6. [14]

    Kalman,R.E.:Anewapproachtolinearfilteringandpredictionproblems.Journal of Basic Engineering82(1), 35–45 (1960)

    1960

  7. [15]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp

    Zhou, K., Yang, Y., Cavallaro, A., Xiang, T.: Omni-scale feature learning for per- son re-identification. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp. 3702–3712 (2019)

    2019

  8. [16]

    In: European Conference on Computer Vision, pp

    Wang, Z., Zheng, L., Liu, Y., Li, Y., Wang, S.: Towards real-time multi-object tracking. In: European Conference on Computer Vision, pp. 107–122 (2020). 15 Springer

    2020

  9. [17]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

    He, S., Luo, H., Wang, P., Wang, F., Li, H., Jiang, W.: Transreid: Transformer- based object re-identification. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 15013–15022 (2021)

    2021

  10. [18]

    bioRxiv (2024) https://doi.org/10.1101/2024.02.07.579241 https://www.biorxiv.org/content/early/2024/11/30/2024.02.07.579241.full.pdf

    Azechi, H., Takahashi, S.: vmtracking: Virtual markers over- come occlusion and crowding in multi-animal pose track- ing. bioRxiv (2024) https://doi.org/10.1101/2024.02.07.579241 https://www.biorxiv.org/content/early/2024/11/30/2024.02.07.579241.full.pdf

    2024 doi

  11. [19]

    bioRxiv (2024) https://doi.org/10.1101/2024.09

    Ichikawa, J., Yamada, M., Fujii, K.: Analysis of coordinated group behavior based on role-sharing: Practical application from an experimental task to a 3-on-3 bas- ketball game as a pilot study. bioRxiv (2024) https://doi.org/10.1101/2024.09. 16.612561

    2024 doi

  12. [20]

    In: Proceedings of the 5th International ACM Workshop on Multimedia Content Analysis in Sports, pp

    Van Zandycke, G., Somers, V., Istasse, M., Don, C.D., Zambrano, D.: Deepsportradar-v1: Computer vision dataset for sports understanding with high quality annotations. In: Proceedings of the 5th International ACM Workshop on Multimedia Content Analysis in Sports, pp. 1–8 (2022)

    2022

  13. [21]

    International journal of computer vision129, 548–578 (2021)

    Luiten, J., Osep, A., Dendorfer, P., Torr, P., Geiger, A., Leal-Taixé, L., Leibe, B.: Hota: A higher order metric for evaluating multi-object tracking. International journal of computer vision129, 548–578 (2021)

    2021

  14. [22]

    arXiv preprint arXiv:2203.14360 (2023)

    Cao, J., Weng, X., Anastasios, A., Kitani, K.: Oc-sort: Reassessing re- identification in multi-object tracking. arXiv preprint arXiv:2203.14360 (2023)

    2023 arXiv

  15. [23]

    arXiv preprint arXiv:2111.14690 (2021)

    Sun, P., Cao, J., Jiang, Y., Yuan, Z., Bai, S., Kitani, K., Luo, P.: Dancetrack: Multi-object tracking in uniform appearance and diverse motion. arXiv preprint arXiv:2111.14690 (2021)

    2021 arXiv

  16. [24]

    In: 2020 25th International Conference on Pattern Recognition (ICPR), pp

    Yang, Y., Zhang, R., Wu, W., Peng, Y., Xu, M.: Multi-camera sports players 3d localization with identification reasoning. In: 2020 25th International Conference on Pattern Recognition (ICPR), pp. 4497–4504 (2021). IEEE

    2021

  17. [25]

    International Journal of Computer Vision129(11), 3069–3087 (2021)

    Zhang, Y., Wang, C., Wang, X., Zeng, W., Liu, W.: Fairmot: On the fairness of detection and re-identification in multiple object tracking. International Journal of Computer Vision129(11), 3069–3087 (2021)

    2021

  18. [26]

    In: 2017 IEEE International Conference on Image Processing (ICIP), pp

    Wojke,N.,Bewley,A.,Paulus,D.:Simpleonlineandrealtimetrackingwithadeep association metric. In: 2017 IEEE International Conference on Image Processing (ICIP), pp. 3645–3649 (2017). IEEE

    2017

  19. [27]

    In: 2016 IEEE International Conference on Image Processing (ICIP), pp

    Bewley, A., Ge, Z., Ott, L., Ramos, F., Upcroft, B.: Simple online and realtime tracking. In: 2016 IEEE International Conference on Image Processing (ICIP), pp. 3464–3468 (2016). https://doi.org/10.1109/ICIP.2016.7533003 16

    2016

  20. [28]

    In: 2018 IEEE Winter Conference on Applications of Computer Vision (WACV), pp

    Wojke, N., Bewley, A.: Deep cosine metric learning for person re-identification. In: 2018 IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 748–756 (2018). https://doi.org/10.1109/WACV.2018.00087 . IEEE

    2018

  21. [29]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    Toshev, A., Szegedy, C.: Deeppose: Human pose estimation via deep neural net- works. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1653–1660 (2014)

    2014

  22. [30]

    Advances in Neural Information Processing Systems 35, 38571–38584 (2022)

    Xu, Y., Zhang, J., Zhang, Q., Tao, D.: Vitpose: Simple vision transformer base- lines for human pose estimation. Advances in Neural Information Processing Systems 35, 38571–38584 (2022)

    2022

  23. [31]

    IEEE transactions on pattern analysis and machine intelligence43(10), 3349–3364 (2020)

    Wang, J., Sun, K., Cheng, T., Jiang, B., Deng, C., Zhao, Y., Liu, D., Mu, Y., Tan, M., Wang, X.,et al.: Deep high-resolution representation learning for visual recog- nition. IEEE transactions on pattern analysis and machine intelligence43(10), 3349–3364 (2020)

    2020

  24. [32]

    Nature Methods19(4), 496–504 (2022)

    Lauer, J., Zhou, M., Ye, S., Menegas, W., Schneider, S., Nath, T., Rahman, M.M., Di Santo, V., Soberanes, D., Feng, G.,et al.: Multi-animal pose estima- tion, identification and tracking with deeplabcut. Nature Methods19(4), 496–504 (2022)

    2022

  25. [33]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    Cao, Z., Simon, T., Wei, S.-E., Sheikh, Y.: Realtime multi-person 2d pose esti- mation using part affinity fields. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 7291–7299 (2017)

    2017

  26. [34]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    Pishchulin, L., Insafutdinov, E., Tang, S., Andres, B., Andriluka, M., Gehler, P.V., Schiele, B.: Deepcut: Joint subset partition and labeling for multi person pose estimation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 4929–4937 (2016)

    2016

  27. [35]

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

    Wang, M., Tighe, J., Modolo, D.: Combining detection and tracking for human pose estimation in videos. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11088–11096 (2020)

    2020

  28. [36]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    Iqbal, U., Milan, A., Gall, J.: Posetrack: Joint multi-person pose estimation and tracking. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 2011–2020 (2017)

    2017

  29. [37]

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

    Raaj, Y., Idrees, H., Hidalgo, G., Sheikh, Y.: Efficient online multi-person 2d pose tracking with recurrent spatio-temporal affinity fields. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4620–4628 (2019)

    2019

  30. [38]

    In: Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 17 pp

    Insafutdinov, E., Andriluka, M., Pishchulin, L., Tang, S., Levinkov, E., Andres, B., Schiele, B.: Arttrack: Articulated multi-person tracking in the wild. In: Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 17 pp. 6457–6465 (2017)

    2017

  31. [39]

    Nature Neuroscience21, 1281–1289 (2018) https://doi.org/ 10.1038/s41593-018-0209-y

    Mathis, A., Mamidanna, P., Cury, K.M., Abe, T., Murthy, V.N., Mathis, M.W., Bethge, M.: Deeplabcut: markerless pose estimation of user-defined body parts with deep learning. Nature Neuroscience21, 1281–1289 (2018) https://doi.org/ 10.1038/s41593-018-0209-y

    2018 doi

  32. [40]

    In: Proceedings of 5-th Berkeley Symposium on Mathematical Statistics and Probability/University of California Press (1967)

    MacQueen, J.: Some methods for classification and analysis of multivariate obser- vations. In: Proceedings of 5-th Berkeley Symposium on Mathematical Statistics and Probability/University of California Press (1967)

    1967

  33. [41]

    In: International Conference on Machine Learning, pp

    Tan, M., Le, Q.: Efficientnet: Rethinking model scaling for convolutional neu- ral networks. In: International Conference on Machine Learning, pp. 6105–6114 (2019). PMLR

    2019

  34. [42]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recogni- tion. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778 (2016)

    2016

  35. [43]

    arXiv preprint arXiv:1704.04861 (2017)

    Howard, A.G.: Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861 (2017)

    2017 arXiv

  36. [44]

    https://github.com/ ultralytics/ultralytics 18

    Jocher, G., Chaurasia, A., Qiu, J.: Ultralytics YOLOv8. https://github.com/ ultralytics/ultralytics 18

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.