Pith. sign in

REVIEW 5 major objections 6 minor 55 references

Ice Hockey Puck Localization Using Contextual Cues

T0 review · 5 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read By feeding a network a picture of only the players, PLUCC boosts single-frame puck detection by 12.3% mean average precision over the best generic baseline.

desk verdict A credible context-cue architecture and a well-motivated metric, but the 12.3% SOTA claim rests on test-set hyperparameter tuning and a comparison set too thin to support it. read the letter →

arxiv 2506.04365 v1 pith:FKFWXU7N submitted 2025-06-04 cs.CV cs.AI

Liam Salass , Jerrin Bright , Amir Nazemi , Yuhao Chen , John Zelek , David Clausi This is my paper
classification cs.CVcs.AI
keywords icehockeypuckdetectioncontextualcuesplayersegmentationgatedfeaturefusionGaussianheatmaprink-spacelocalizationerrorhomographyevaluationsingle-frame
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 player behavior is a strong, learnable cue for locating a hockey puck in broadcast video: players consistently turn their bodies and gaze toward the puck, and a network that is explicitly shown a segmented player image can use that cue to localize the puck where appearance alone fails. The proposed system, PLUCC, processes the full-resolution frame and a half-resolution image containing only the segmented players through two encoders, fuses their multi-scale features with a learned gating mechanism, and outputs a Gaussian heatmap of puck position. On the proprietary VIP-PuckDataset test set, PLUCC reports a 12.3% mean average precision improvement over the strongest generic detector tested (Faster-RCNN), and a new rink-space metric, RSLE, shows its detections land within one puck radius more than twice as often as that baseline. If correct, this means single-frame, context-driven detection is a viable path to accurate puck tracking without multi-frame temporal methods or expensive player-tracking hardware.

What carries the argument

The load-bearing mechanism is a gated feature-fusion decoder operating on two encoders: a ResNet-152 feature pyramid that processes the full-resolution RGB frame, and a context encoder that processes a half-resolution RGB image containing only the segmented players, produced by a frozen player detector followed by an off-the-shelf segmentation model. At each fusion stage the decoder concatenates corresponding-scale features from the two encoders and applies a static per-channel gating vector (a sigmoid over learned weights) before a convolution block, letting the context selectively amplify or suppress visual channels. Training uses Gaussian heatmap labels centered on the puck with a KL-divergence loss, and a 1% context-driven dropout of the raw frame pushes the network to lean on player cues. For evaluation, RSLE computes a per-frame homography into standardized rink coordinates so that pixel distances are converted to physical meters, removing the perspective bias of broadcast camera angles.

What would settle it

Replace the player-segmentation context image in the PLUCC pipeline with a deliberately uninformative version (e.g., random masks or masks of players shifted to the opposite side of the rink) and measure mAP on the VIP-PuckDataset test set; if the score does not drop materially, the reported gains do not come from player context. A complementary test is to compare PLUCC's accuracy on frames where the puck is far from any player against frames where players surround the puck, since the contextual cue should be weakest in the former.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that player context is a learnable prior strong enough to carry puck localization: PLUCC, which fuses RGB player segmentations with standard image features in a gated multi-scale decoder, outperforms every baseline detector at every distance threshold, with mean average precision of 83.5 versus 71.3 for Faster-RCNN and 79.6 for the context-free FCN-ResNet152. The largest gains appear at the strictest threshold (AP5 of 82.2 versus 69.6), which the authors attribute to the context encoder implicitly capturing player orientation and position. An ablation supports this interpretation: a network trained only on the player-segmentation image predicts heatmaps that peak in the direction of players' gaze, even though its absolute accuracy is low, and a 1% training dropout of the raw frame raises AP5 by 3.5 points and lowers average rink-space error from 1.36 to 1.05 meters. The paper also introduces RSLE, a homography-based metric that projects image detections into NHL rink coordinates, and under that metric PLUCC achieves 43.6% average precision within one puck radius compared with 18.6% for the best generic baseline.

Load-bearing premise

The claim depends on the pretrained player detector and segmenter producing masks that reliably encode player position and body direction under the exact broadcast conditions in the test set; if these upstream models fail on a meaningful fraction of frames, the context encoder receives noise and the reported advantage could shrink.

Editorial extensions

If this is right

  • PLUCC produces a detection per frame with no temporal input, so it can be deployed in streaming settings and will not inherit error propagation from multi-frame tracking methods.
  • Because the output is a Gaussian heatmap, the detections plug directly into the heatmap-based sports-ball tracking pipeline that motivated the label design.
  • The 1% context-driven dropout is a cheap training change that buys 3.5 points of AP5 and 0.31 meters of rink-space accuracy, making the model more robust to occlusion and visually masked pucks.
  • The RSLE metric gives hockey analytics a way to compare detectors in physical rink coordinates, which the authors argue is fairer than pixel-space distances that are distorted by camera perspective.

Reading between the lines

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

  • Editorial inference: the same context-encoder design may transfer to other broadcast sports in which a small object's position is implied by player orientation and spacing, such as lacrosse or rugby, but the reported 12-point gain should be expected to shrink if players do not reliably orient toward the object.
  • Editorial inference: the paper does not report the upstream player detector's or segmenter's accuracy on this dataset; using ground-truth player masks instead of the frozen pipeline would reveal how much of the measured gain comes from the context signal itself rather than from good upstream segmentation.
  • Editorial inference: RSLE's sensitivity to homography estimation error is not quantified; a natural test is to perturb the per-frame homography matrices and measure how much APr and RSLEavg change.
  • Editorial inference: at 6.03 FPS including preprocessing, the frozen player detector and segmenter dominate runtime; swapping them for faster lightweight models could make the system real-time, with unknown but testable cost to the context signal's usefulness.
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

5 major / 6 minor

Summary. The paper proposes PLUCC, a single-frame heatmap-based puck detector that fuses standard RGB features with an RGB player-segmentation context image through a gated decoder, and introduces RSLE, a homography-based rink-space evaluation metric. The method is evaluated on a proprietary VIP-PuckDataset against Faster-RCNN, YOLOv5, and an FCN-ResNet152 baseline, with reported gains of 12.3% in mAP_tau over Faster-RCNN and large improvements in rink-space AP. The core hypothesis, that player position and gaze provide a useful puck prior, is plausible, and the match-disjoint train/test split is a strength. However, the central SOTA claim is not yet supported because the key Gaussian sigma hyperparameter is selected on the test set, the comparison omits all puck-specific methods discussed in the paper, and the homography underlying RSLE is not validated.

Significance. If the results hold, the main contribution is evidence that pretrained player segmentation, encoded as RGB context, improves single-frame puck localization under challenging broadcast conditions, and the RSLE metric would be a useful complement to pixel-space AP. Strengths include the match-disjoint split, the explicit ablation of the Gaussian target, and the attempt to address perspective bias with a rink-space metric. However, the reported effect sizes are not yet credible because the key hyperparameter is selected on the test set, the comparison omits puck-specific methods, and the homography basis of RSLE is unvalidated. The architecture is a reasonable fusion of existing components, and the application is well motivated, but the paper currently overstates the strength of the evidence.

major comments (5)
  1. [Section 4.6.2 and Table 1] The sigma value used for the main PLUCC results is selected on the test set. The text states that the sigma ablation was evaluated 'on the test set' and Figure 6 selects sigma=5, while Table 1 shows the choice is material: PLUCC sigma=5 gives mAP_tau 83.5 and AP5 82.2, whereas PLUCC sigma=15 gives mAP_tau 81.6 and AP5 76.2. The baselines were not given comparable test-set tuning, so the headline 12.3% mAP_tau gain over Faster-RCNN and the 7.4% AP5 context-encoder gain may reflect selection on the evaluation set rather than a genuine method advantage. Please select sigma on validation and report both tuned and untuned results.
  2. [Section 2.1 and Table 1] The comparison omits all puck-specific methods discussed in the paper, specifically PuckNet [49], Yang [55], Li et al. [24], Sarkhoosh et al. [37], and Pidaparthy et al. [29]. The abstract and Section 5 call PLUCC 'state-of-the-art' and claim it surpasses 'previous baseline methods', but only generic object detectors and one heatmap FCN are evaluated. Please add these methods where they can be run on the same data, or clearly state and justify why they are excluded; otherwise the SOTA claim is not supported.
  3. [Section 4.4.2 and Table 2] The RSLE comparison relies on per-frame homographies from Shang et al. [39], but the paper reports no validation of those homography estimates on VIP-PuckDataset. If H is inaccurate, the reported AP_r and RSLEavg values are not meaningful and the abstract's 25% RSLE improvement is unverifiable. Please report homography registration error on the test set and evaluate how sensitive RSLE is to homography errors.
  4. [All experiments (Tables 1-4)] No error bars or multiple-seed runs are reported. Differences such as PLUCC mAP_tau 83.5 vs FCN-ResNet152 79.6 and AP_r 43.59 vs 41.85 are reported as point estimates; without variance, the claimed 3.9% context-encoder gain and 1.74% rink-space gain cannot be assessed. Please provide mean and standard deviation over at least three seeds and a significance test for the central comparisons.
  5. [Section 3.1.2 and Table 3] The context encoder relies on a pretrained player detector followed by SAM 2 segmentation, but the paper does not report the accuracy of these upstream models on VIP-PuckDataset. Table 3 shows that a network trained only on context images achieves only 6.0 mAP_tau, so the claimed benefit of context depends on the reliability of the frozen upstream masks. Please report the player detector and segmentation accuracy on the test set, and include an ablation that perturbs or corrupts the context masks to establish robustness.
minor comments (6)
  1. [Abstract and Figure 1 caption] There are typos: 'F or evaluation' in the abstract and 'do to similar colours' in the Figure 1 caption.
  2. [Equation (8)] The text calls sigma the variance of the Gaussian label, but the exponent uses 2*sigma^2, which means sigma is actually the standard deviation; please clarify.
  3. [Equation (12)] The variable list repeats 'xrink' twice ('xrink and xrink'); the second should be 'yrink'.
  4. [Section 3.1.2] The 'pretrained detector' used to generate player bounding boxes is never cited or named; please specify the detector architecture and checkpoint.
  5. [Section 4.2] The text says 'using billinear interpolation' in the decoder description; it should be 'bilinear'.
  6. [Abstract and Section 4.5] The abstract reports a 12.2% average precision improvement while Section 4.5 reports 12.3%; please harmonize these numbers.

Circularity Check

0 steps flagged · score 1.0 of 10

No constructional circularity: the reported gains are empirical, though the σ=5 choice is made on the test set, which is a correctness concern rather than a circularity.

full rationale

PLUCC's derivation chain is self-contained: the predicted puck center is obtained by argmax over a softmax-normalized heatmap, trained with KL divergence against Gaussian labels whose centers come from ground-truth bounding boxes (Sec. 3.2, Eqs. 6–8). No output quantity is defined in terms of the model's own predictions, and no fitted parameter is renamed as a prediction. The context encoder is an external frozen detection-plus-segmentation pipeline (Sec. 3.1.2), and the paper's own ablation shows context alone yields only 6.0 mAP (Table 3), so the context signal is not a definitional stand-in for the final result. The only apparent self-citation, the homography method of Shang et al. [39] used for RSLE, is an externally published geometric registration method and is used only to transform coordinates; it is not load-bearing in the sense of importing a uniqueness claim or defining the detection output. The genuine weakness is that the Gaussian variance σ was selected by evaluating models on the test set (Sec. 4.6.2: 'we conducted an ablation study by training models with different σ values... on the test set, where the model trained with σ = 5 achieved the best performance'), which can inflate PLUCC's test-set numbers relative to baselines that were not tuned on that same test set. This is a statistical soundness issue, not circularity: the reported margin is not mathematically forced by construction, and the model's advantage remains an empirical claim, albeit one whose magnitude may be optimistically biased.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim rests on two fitted hyperparameters (sigma, pdrop) and three domain assumptions: the player-gaze correlation, the reliability of the pretrained player detector and segmenter, and the accuracy of the homography used for RSLE. No invented physical entities are introduced. The main methodological weakness is that sigma and pdrop were selected using test-set performance, which is scored under soundness and red flags rather than as circularity.

free parameters (2)
  • Gaussian heatmap variance sigma = 5.0
    Chosen by ablation on the test set (Section 4.6.2); controls the sharpness of the Gaussian heatmap target and affects AP at tight thresholds.
  • Context-driven dropout probability pdrop = 1%
    Set by comparing pdrop=0 vs pdrop=1 on the test set (Section 4.6.3, Table 4); forces the model to rely on contextual cues.
assumptions (3)
  • domain assumption Players consistently turn their bodies and direct their gaze toward the puck.
    Stated in the abstract and Section 3.1.2 as motivation; if this correlation is weak, the context encoder provides little signal. The paper's own ablation shows context-only localization is poor (mAP 6.0), so the cue is weak in isolation.
  • domain assumption Pretrained player detector and SAM 2 segmentation generalize to broadcast hockey frames.
    Used to generate CRGB in Section 3.1.2; no accuracy metrics are reported on the test distribution, so the reliability of the context input is assumed.
  • domain assumption The homography estimated by Shang et al. [39] is accurate per frame.
    Used in Section 4.4.2 to compute RSLE; no homography error is reported, so the metric's validity is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Ice Hockey Puck Localization Using Contextual Cues." pith.science (2026). https://pith.science/paper/FKFWXU7N

@misc{pith2026250604365,
  author       = {Pith},
  title        = {Pith review of: Ice Hockey Puck Localization Using Contextual Cues},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FKFWXU7N}},
  note         = {Machine review of arXiv:2506.04365}
}
read the original abstract

Puck detection in ice hockey broadcast videos poses significant challenges due to the puck's small size, frequent occlusions, motion blur, broadcast artifacts, and scale inconsistencies due to varying camera zoom and broadcast camera viewpoints. Prior works focus on appearance-based or motion-based cues of the puck without explicitly modelling the cues derived from player behaviour. Players consistently turn their bodies and direct their gaze toward the puck. Motivated by this strong contextual cue, we propose Puck Localization Using Contextual Cues (PLUCC), a novel approach for scale-aware and context-driven single-frame puck detections. PLUCC consists of three components: (a) a contextual encoder, which utilizes player orientations and positioning as helpful priors; (b) a feature pyramid encoder, which extracts multiscale features from the dual encoders; and (c) a gating decoder that combines latent features with a channel gating mechanism. For evaluation, in addition to standard average precision, we propose Rink Space Localization Error (RSLE), a scale-invariant homography-based metric for removing perspective bias from rink space evaluation. The experimental results of PLUCC on the PuckDataset dataset demonstrated state-of-the-art detection performance, surpassing previous baseline methods by an average precision improvement of 12.2% and RSLE average precision of 25%. Our research demonstrates the critical role of contextual understanding in improving puck detection performance, with broad implications for automated sports analysis.

Figures

Figures reproduced from arXiv: 2506.04365 by the authors.

Figure 1
Figure 1. Test set examples of challenges faced in automated puck [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Left image (a) shows context image generation pipeline, a combination of detection and segmentation models frozen in training. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Qualitative results of PLUCC with expanded Gaussian overlays. Sub-figure (a) demonstrates robust detection under heavy puck blurring conditions, (b) shows model resistance to out-of-distribution broadcast artifacts, (c) highlights a correct puck detection even under full occlusion from a hockey stick, and (d) showcases the model’s ability to predict the puck location when it is on a yellow-line, where the contrast b… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Visualization of (a) heatmap detection transformed to [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 6
Figure 6. Figure 6: Comparison of performance of different models trained [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

55 extracted references · 49 canonical work pages

  1. [49]

    Puck localization and multi-task event recognition in broad- cast hockey videos

    Kanav Vats, Mehrnaz Fani, David A Clausi, and John Zelek. Puck localization and multi-task event recognition in broad- cast hockey videos. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition , pages 4567–4575, 2021. 1, 2, 3

    work page 2021

  2. [55]

    Where is the puck? tiny and fast-moving object detection in videos

    Xionghao Yang. Where is the puck? tiny and fast-moving object detection in videos. 2021. 2, 5 10

    work page 2021

  3. [24]

    Ice hockey puck track- ing through broadcast video

    Muyu Li, Henan Hu, and Hong Yan. Ice hockey puck track- ing through broadcast video. Neurocomputing, 551:126484,

    work page

  4. [37]

    Riegler, and P ˚al Halvorsen

    Mehdi Houshmand Sarkhoosh, Sayed Mohammad Majidi Dorcheh, Cise Midoglu, Saeed Shafiee Sabet, Tomas Kupka, Dag Johansen, Michael A. Riegler, and P ˚al Halvorsen. AI- Based Cropping of Ice Hockey Videos for Different Social Media Representations. IEEE Access , 12:118227–118249,

    work page

  5. [29]

    Keep your eye on the puck: Automatic hockey videography

    Hemanth Pidaparthy and James H Elder. Keep your eye on the puck: Automatic hockey videography. In Winter Confer- ence on Applications in Computer Vision (WACV) , 2019. 2, 3

    work page 2019

  6. [39]

    Jia Cheng Shang, Yuhao Chen, Mohammad Javad Shafiee, and David A. Clausi. Rink-agnostic hockey rink registration. In Proceedings of the 6th International Workshop on Multi- media Content Analysis in Sports , page 73–81, New York, NY , USA, 2023. Association for Computing Machinery. 7

    work page 2023

  7. [1]

    Digital camera work for soccer video production with event recogni- tion and accurate ball tracking by switching search method

    Yasuo Ariki, Tetsuya Takiguchi, and Kazuki Yano. Digital camera work for soccer video production with event recogni- tion and accurate ball tracking by switching search method. In 2008 IEEE International Conference on Multimedia and Expo, pages 889–892, 2008. 2

    work page 2008

  8. [2]

    Domain-guided masked autoencoders for unique player identification

    Bavesh Balaji, Jerrin Bright, Sirisha Rambhatla, Yuhao Chen, Alexander Wong, John Zelek, and David A Clausi. Domain-guided masked autoencoders for unique player identification. 1

    work page

Show all 55 references
  1. [3]

    Seeing beyond the crop: Using language priors for out-of-bounding box key- point prediction

    Bavesh Balaji, Jerrin Bright, Yuhao Chen, Sirisha Ramb- hatla, John Zelek, and David Clausi. Seeing beyond the crop: Using language priors for out-of-bounding box key- point prediction. Advances in Neural Information Processing Systems, 37:102897–102918, 2024. 1

    2024

  2. [4]

    Aspogamo: Automated sports games analysis models

    Michael Beetz, Nico von Hoyningen-Huene, Bernhard Kirchlechner, Suat Gedikli, Francisco Siles, Murat Durus, and Martin Lames. Aspogamo: Automated sports games analysis models. Int. J. Comput. Sci. Sport , 8, 2009. 2

    2009

  3. [5]

    Nhl technology showcase gives glimpse of future of hockey consumption

    Amalie Benjamin. Nhl technology showcase gives glimpse of future of hockey consumption. https://www.nhl.com/news/nhl-technology- showcase - gives - glimpse - of - future - of - hockey-consumption-342952272, 2023. Accessed: 2025-03-10. 8

    2023

  4. [6]

    2d trajectory- based position estimation and tracking of a ball in a basket- ball video

    Bodhisattwa Chakraborty and Sukadev Meher. 2d trajectory- based position estimation and tracking of a ball in a basket- ball video. 2011. 2

    2011

  5. [7]

    Real-time po- sition estimation and tracking of a basketball

    Bodhisattwa Chakraborty and Sukadev Meher. Real-time po- sition estimation and tracking of a basketball. 2012 IEEE International Conference on Signal Processing, Computing and Control, pages 1–6, 2012

    2012

  6. [8]

    A trajectory- based ball detection and tracking system with applications to shooting angle and velocity estimation in basketball videos

    Bodhisattwa Chakraborty and Sukadev Meher. A trajectory- based ball detection and tracking system with applications to shooting angle and velocity estimation in basketball videos. In 2013 Annual IEEE India Conference (INDICON) , pages 1–6, 2013. 2

    2013

  7. [9]

    Physics-based ball tracking in volleyball videos with its ap- plications to set type recognition and action detection

    Hua-Tsung Chen, Hsuan-Shen Chen, and Suh-Yin Lee. Physics-based ball tracking in volleyball videos with its ap- plications to set type recognition and action detection. In 2007 IEEE International Conference on Acoustics, Speech and Signal Processing - ICASSP ’07 , pages I–1097–I–1100,

    2007

  8. [10]

    Ball tracking and 3d trajectory approximation with ap- plications to tactics analysis from single-camera volleyball sequences

    Hua-Tsung Chen, Wen-Jiin Tsai, Suh-Yin Lee, and Jen-Yu Yu. Ball tracking and 3d trajectory approximation with ap- plications to tactics analysis from single-camera volleyball sequences. Multimedia Tools Appl., 60(3):641–667, 2012. 2

    2012

  9. [11]

    Transformer-based context condensation for boosting feature pyramids in object detection

    Zhe Chen, Jing Zhang, Yufei Xu, and Dacheng Tao. Transformer-based context condensation for boosting feature pyramids in object detection. International Journal of Com- puter Vision, 131(10):2738–2756, 2023. 3

    2023

  10. [12]

    Particle filter with ball size adaptive tracking window and ball feature likelihood model for ball’s 3d position tracking in volleyball analysis

    Xina Cheng, Xizhou Zhuang, Yuan Wang, Masaaki Honda, and Takeshi Ikenaga. Particle filter with ball size adaptive tracking window and ball feature likelihood model for ball’s 3d position tracking in volleyball analysis. pages 203–211,

  11. [13]

    Anti-occlusion observation model and automatic recovery for multi-view ball tracking in sports analysis

    Xina Cheng, Masaaki Honda, Norikazu Ikoma, and Takeshi Ikenaga. Anti-occlusion observation model and automatic recovery for multi-view ball tracking in sports analysis. In 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1501–1505, 2016. 2

    2016

  12. [14]

    Probabilistic tracking of the soccer ball

    Kyuhyoung Choi and Yongduek Seo. Probabilistic tracking of the soccer ball. pages 50–60, 2004. 2

    2004

  13. [15]

    Tracking soccer ball in tv broadcast video

    Kyuhyoung Choi and Yongduek Seo. Tracking soccer ball in tv broadcast video. pages 661–668, 2005. 2

    2005

  14. [16]

    Automatic determination of puck possession and location in broadcast hockey video

    Xin Duan. Automatic determination of puck possession and location in broadcast hockey video . PhD thesis, University of British Columbia, 2011. 1

    2011

  15. [17]

    Object detection via a multi-region & semantic segmentation-aware CNN model

    Spyros Gidaris and Nikos Komodakis. Object detection via a multi-region & semantic segmentation-aware CNN model. CoRR, abs/1505.01749, 2015. 3

    2015 arXiv

  16. [18]

    Ball tracking based on multiscale feature enhancement and cooperative trajec- tory matching

    Xiao Han, Qi Wang, and Yongbin Wang. Ball tracking based on multiscale feature enhancement and cooperative trajec- tory matching. Applied Sciences, 14:1376, 2024. 2

    2024

  17. [19]

    Deep residual learning for image recognition, 2015

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition, 2015. 3

    2015

  18. [20]

    Mask r-cnn, 2018

    Kaiming He, Georgia Gkioxari, Piotr Doll ´ar, and Ross Gir- shick. Mask r-cnn, 2018. 2

    2018

  19. [21]

    Squeeze-and-excitation networks, 2019

    Jie Hu, Li Shen, Samuel Albanie, Gang Sun, and Enhua Wu. Squeeze-and-excitation networks, 2019. 4, 8

    2019

  20. [22]

    Nhl debuts new tracking tech from hawk- eye

    Joe Lemire. Nhl debuts new tracking tech from hawk- eye. https : / / www . sportsbusinessjournal . com/Daily/Morning- Buzz/2023/03/31/nhl- tech- showcase- hawk- eye- technology/, 2023. Accessed: 2025-03-10. 8

    2023

  21. [23]

    Mosca, Paolo Spagnolo, and Pier Luigi Mazzeo

    Marco Leo, N. Mosca, Paolo Spagnolo, and Pier Luigi Mazzeo. A semi-automatic system for ground truth gener- ation of soccer video sequences. pages 559 – 564, 2009. 2

    2009

  22. [25]

    Content adaptive and error propagation aware deep video compression

    Guo Lu, Chunlei Cai, Xiaoyun Zhang, Li Chen, Wanli Ouyang, Dong Xu, and Zhiyong Gao. Content adaptive and error propagation aware deep video compression. In Com- puter Vision–ECCV 2020: 16th European Conference, Glas- gow, UK, August 23–28, 2020, Proceedings, Part II 16, pages...

    2020

  23. [26]

    Hockey puck math: For radius, di- ameter, area, circumference, and volume

    Brian Nemeth. Hockey puck math: For radius, di- ameter, area, circumference, and volume. https : / / bleacherreport . com / articles / 412092 - hockey - puck - math - for - radius - diameter - area - circumference - and - volume, 2018. Ac- cessed: 2025-03-10. 7

    2018

  24. [27]

    Leveraging player tracking for event detection in ice hockey

    Ken Nsiempba, Amir Nazemi, David Clausi, and John Zelek. Leveraging player tracking for event detection in ice hockey. Journal of Computational Vision and Imaging Systems , 10 (1):69–74, 2024. 1 9

    2024

  25. [28]

    Balaji, bavesh

    Language Guided Out of-Bounding Box Pose Estimation for Robust Ice Hockey Analysis. Balaji, bavesh. Master’s thesis, University of Waterloo, 2024. 3

    2024

  26. [31]

    Multi player tracking in ice hockey with homographic projections

    Harish Prakash, Jia Cheng Shang, Ken M Nsiempba, Yuhao Chen, David A Clausi, and John S Zelek. Multi player tracking in ice hockey with homographic projections. arXiv preprint arXiv:2405.13397, 2024. 1

    2024 arXiv

  27. [32]

    Sam 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman R¨adle, Chloe Rolland, Laura Gustafson, Eric Mintun, Junt- ing Pan, Kalyan Vasudev Alwala, Nicolas Carion, Chao- Yuan Wu, Ross Girshick, Piotr Doll´ar, and Christoph Feic...

    2024 arXiv

  28. [33]

    You only look once: Unified, real-time object de- tection

    J Redmon. You only look once: Unified, real-time object de- tection. In Proceedings of the IEEE conference on computer vision and pattern recognition, 2016. 5, 6, 7

    2016

  29. [34]

    Yolov3: An incremental improvement, 2018

    Joseph Redmon and Ali Farhadi. Yolov3: An incremental improvement, 2018. 2

    2018

  30. [35]

    Jones, and Ming Xu

    Jinchang Ren, James Orwell, Graeme A. Jones, and Ming Xu. Tracking the soccer ball using multiple fixed cam- eras. Computer Vision and Image Understanding , 113(5): 633–642, 2009. Computer Vision Based Analysis in Sport Environments. 2

    2009

  31. [36]

    Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks, 2016

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks, 2016. arXiv:1506.01497 [cs]. 2, 3, 5, 6, 7

    2016 arXiv

  32. [38]

    Goalienet: A multi-stage network for joint goalie, equip- ment, and net pose estimation in ice hockey

    Marjan Shahi, David Clausi, and Alexander Wong. Goalienet: A multi-stage network for joint goalie, equip- ment, and net pose estimation in ice hockey. arXiv preprint arXiv:2306.15853, 2023. 1

    2023 arXiv

  33. [40]

    Hs-fpn: High frequency and spatial perception fpn for tiny object detec- tion, 2024

    Zican Shi, Jing Hu, Jie Ren, Hengkang Ye, Xuyang Yuan, Yan Ouyang, Jia He, Bo Ji, and Junyu Guo. Hs-fpn: High frequency and spatial perception fpn for tiny object detec- tion, 2024. 3

    2024

  34. [41]

    What data-driven hockey software and puck data can do for coaches and players

    KINEXON Sports. What data-driven hockey software and puck data can do for coaches and players. https : //kinexon- sports.com/blog/data- driven- hockey-software/ , 2025. Accessed: 2025-03-10. 1, 8

    2025

  35. [42]

    Size of hockey rinks: Why the us rink is smaller than the eu rink

    Athletica Sport Systems. Size of hockey rinks: Why the us rink is smaller than the eu rink. https://www. athletica.com/size- of- hockey- rinks/. Ac- cessed: 2025-03-11. 7

    2025

  36. [43]

    Widely Applicable Strong Baseline for Sports Ball Detection and Tracking, 2023

    Shuhei Tarashima, Muhammad Abdul Haq, Yushan Wang, and Norio Tagawa. Widely Applicable Strong Baseline for Sports Ball Detection and Tracking, 2023. arXiv:2311.05237 [cs]. 3, 5, 8

    2023 arXiv

  37. [44]

    Group event recognition in ice hockey

    Sijia Tian. Group event recognition in ice hockey. PhD thesis, University of British Columbia, 2018. 1

    2018

  38. [45]

    Clas- sification of puck possession events in ice hockey

    Moumita Roy Tora, Jianhui Chen, and James J Little. Clas- sification of puck possession events in ice hockey. In 2017 IEEE conference on computer vision and pattern recognition workshops (CVPRW), pages 147–154. IEEE, 2017. 1

    2017

  39. [46]

    Yolov8: A novel object detection algorithm with enhanced performance and robust- ness

    Rejin Varghese and Sambath M. Yolov8: A novel object detection algorithm with enhanced performance and robust- ness. In 2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS) , pages 1–6, 2024. 2

    2024

  40. [47]

    D-pose: Depth as an intermediate representation for 3d human pose and shape estimation, 2024

    Nikolaos Vasilikopoulos, Drosakis Drosakis, and Antonis Argyros. D-pose: Depth as an intermediate representation for 3d human pose and shape estimation, 2024. 7

    2024

  41. [48]

    Multi-task learning for jersey number recognition in ice hockey

    Kanav Vats, Mehrnaz Fani, David A Clausi, and John Zelek. Multi-task learning for jersey number recognition in ice hockey. In Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports , pages 11–15, 2021. 1

    2021

  42. [50]

    Player tracking and identification in ice hockey

    Kanav Vats, Pascale Walters, Mehrnaz Fani, David A Clausi, and John S Zelek. Player tracking and identification in ice hockey. Expert systems with applications , 213:119250,

  43. [51]

    Adaptation of yolov7 and yolov7 tiny for soccer- ball multi-detection with deepsort for tracking by semi- supervised system

    Jorge Armando Vicente-Mart´ınez, Mois´es-Vicente M´arquez- Olivera, Abraham Garc´ıa-Aliaga, and Viridiana Hern´andez- Herrera. Adaptation of yolov7 and yolov7 tiny for soccer- ball multi-detection with deepsort for tracking by semi- supervised system. Sensors (Basel, Switzerla...

    2023

  44. [52]

    A review of small object detection based on deep learning

    Wei Wei, Yu Cheng, Jiafeng He, and Xiyue Zhu. A review of small object detection based on deep learning. Neural Computing and Applications, 36(12):6283–6303, 2024. 3

    2024

  45. [53]

    V ote from the Center: 6 DoF Pose Estimation in RGB-D Images by Radial Keypoint V oting, 2022

    Yangzheng Wu, Mohsen Zand, Ali Etemad, and Michael Greenspan. V ote from the Center: 6 DoF Pose Estimation in RGB-D Images by Radial Keypoint V oting, 2022. 3, 5, 6, 7, 8

    2022

  46. [54]

    Nhl brings advanced puck tracking stats to public, 2023

    Greg Wyshynski. Nhl brings advanced puck tracking stats to public, 2023. 1

    2023

  47. [2024]

    Conference Name: IEEE Access. 2, 5

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.