Pith. Sign in

REVIEW

Event-based Gaze Control System for Accurate Real-time Spin Estimation in Professional Ball Games

PaperPDF from arXiv

data files

ReviewThe objections and verdict, as JSON Paper breakdownThe authors' claims, conditions and measurements, as JSON Signed receiptProof this review has not been altered
X Bluesky LinkedIn Reddit Hacker News
Yunpu Hu, Fabian Schilling, Valentina Cavinato, Asude Aydin, Agis Politis, Ricardo Tapiador Morales, Kirk Y.W. Scheper, Peter Dürr
show 1 more author
Naoya Takahashi

Computer Vision and Pattern Recognition · Submitted 25 Jun 2026 · 46 references

Pith's verdict: The paper delivers a working event-camera plus active-optics pipeline for real-time spin on table tennis balls, but the online accuracy rests on pseudo-labels and the tracking continuity numbers are missing.

T0 · AI review

How this review was made: An AI referee (grok-4.3) read the paper on July 1, 2026 and answered a fixed list of questions about it. T0 is the starting level every paper gets. The levels describe how deeply Pith has checked the paper, never the authors.

  1. T0 AI review An AI read the paper and answered a fixed list of questions.
  2. T1 Audit Automatic checks that the references and files the paper relies on exist and match.
  3. T2 Compute The paper's code or calculations re-run with the same results.
  4. T3 Formal statement The claims written out in a formal math language a computer can read.
  5. T4 Proof Those claims proved, and the proof checked by computer.

How Pith checks papers →

· 3 major objections · 1 minor · Ask about this paper · Read the full review →
Challenge this review Comment (0) Read on arXiv ↗
Cite ▾
arXiv ID2606.26780 (version 2)
Pith IDK53HKZLU, Pith's permanent record of this paper and its review
CitationAgis Politis, Asude Aydin, Fabian Schilling, Kirk Y.W. Scheper, Naoya Takahashi, Peter D\"urr, Ricardo Tapiador Morales, Valentina Cavinato, Yunpu Hu. "Event-based Gaze Control System for Accurate Real-time Spin Estimation in Professional Ball Games." arXiv:2606.26780 (2026). https://pith.science/paper/K53HKZLU
Cite this reviewBibTeX for Pith's review of this paper
Claim paper

Why Pith says this

Highlight any text to comment on it.

Referee report

How Pith checks papers → · Full audit of this paper → · Review history (2 revisions) →

The paper

The paper claims

An event-based active vision system tracks unmodified balls and estimates their spin in real time during professional matches.

The reading

The paper introduces a system using an event camera, high-speed pan-tilt mirrors, and a focus-tunable lens to track and measure ball spin despite high speeds and small size. A hybrid approach combines 2D event-based detection for centering with 3D positions from an external localization system for re-initialization and continuous focus. The offline s-CMax method performs contrast maximization on the sphere and reaches state-of-the-art accuracy on static balls across sports. An online version uses an uncertainty-aware CNN trained on offline labels plus GPU-accelerated refinement to deliver real-time performance in table tennis.

What carries the argument

s-CMax contrast maximization performed on the sphere, paired with hybrid 2D event-based detection and 3D re-initialization for active gaze control.

Extended reading notes

Core claim

The offline s-CMax method achieves mean magnitude and axis errors of 1.2 percent and 1.5 degrees on static balls in table tennis, baseball, tennis, and golf. The online method, tested in professional table tennis matches with a three-view setup, shows 8.8 percent magnitude and 6.4 degrees axis mismatch relative to the offline method while running at 3 ms latency and 750 Hz throughput.

Load-bearing premise

The external 3D ball localization system supplies positions accurate enough to re-initialize and sustain tracking without frequent target loss during fast professional play.

Editorial extensions

If this is right

  • The offline method supplies high-accuracy spin data for analysis in table tennis, baseball, tennis, and golf without ball modifications.
  • The online method supports real-time spin feedback at 750 Hz during live professional table tennis matches.
  • The active vision hardware maintains focus and centering on fast-moving balls with 3 ms latency.
  • Pseudo-ground-truth labels from the offline method enable training of real-time neural networks for spin estimation.

Reading between the lines

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

  • The same hardware and tracking pipeline could be adapted to other high-speed ball sports by swapping the external localization module.
  • Combining the spin estimates with existing physics-based trajectory models would allow more precise prediction of ball paths.
  • The 750 Hz throughput opens the possibility of feeding spin data into automated coaching systems or broadcast graphics in real time.
Original abstract

Spin plays a crucial role in many ball sports due to its effect on the trajectory of the ball. Vision-based estimation of the ball's spin during a game with conventional cameras is challenging due to the ball's small size, high speed, and fast rotation. To address these challenges, we propose an event-based active vision system that can track unmodified balls and measure their spin in real time. The system consists of an event camera for its high temporal resolution and minimal motion blur, high-speed pan/tilt galvanometer mirrors to keep the ball in the field of view, and a low-latency focus-tunable telephoto lens to increase the spatial resolution on the ball and keep it in focus. To track the ball, we use a hybrid approach that combines 2D event-based detection for centering and 3D positions from a ball localization system for re-initialization. For high-accuracy spin estimation, we propose an offline method that performs contrast maximization on the sphere (s-CMax). This method achieves state-of-the-art accuracy on static balls across multiple sports (table tennis, baseball, tennis, and golf), with mean magnitude and axis errors of 1.2% and 1.5 degrees, respectively. We then develop a low-latency online method for table tennis as a case study in real-time applications. This method uses an uncertainty-aware convolutional neural network trained on pseudo-ground-truth spin labels from the offline approach, combined with a GPU-accelerated batch implementation of contrast maximization for refinement. We demonstrate reliable tracking and spin estimation with a three-view setup during professional table tennis matches, with high accuracy (8.8% magnitude and 6.4 degrees axis mismatch w.r.t. the offline method), 3 ms latency, and 750 Hz throughput.

Figures
Figure 1. The gaze control system (GCS) actively tracks the ball and measures its spin in ball games. 1 Introduction Spin plays a key role in many ball sports due to its effect on the ball’s flight and contact dynamics. Athletes routinely apply spin to control the ball’s flight path, deceive opponents, or gain a tactical advantage. Accurate spin estimation is thus of vital importance for performance analysis, sports statistics, and competitive robotic gameplay [5]. For example, in table tennis, rallies can feature balls spin￾ning at magnitudes of up to 9 krpm while flying at speeds of up to…
Figure 2. Overview of the information flow in the GCS. the event stream to set velocity constraints. However, all existing event-based methods employ static, wide-angle cameras, which severely limits the spatial resolution on the ball and conflates translational motion with spin, a funda￾mental challenge for in-flight estimation. Our work addresses both limitations through active tracking with a telephoto lens, proposes a spherical formulation of CMax for more accurate rotation estimation, and develops a real-time pipeline combining an uncertainty-aware CNN with GPU-accelerated CMax…
Figure 3. The effect of different ball types on the s-CMax magnitude and axis error. The bottom row shows a time surface representation of the different balls when their pattern is visible, spinning at 1 krpm. We recorded a dataset with different balls in a sidespin configuration with a total of 19 raw event files, 9 files with the table tennis ball from 1-9 krpm, 4 files for the golf and tennis ball from 1-4 krpm, as well as 2 files for the baseball from 1-2 krpm. We additionally recorded 108 table tennis specific raw event files with three different spin types (topspin, sidespin, and…
Figure 4. CNN magnitude and axis errors on the held-out test set for different uncertainty percentiles and magnitude ranges. 5 Conclusion We presented an event-based active vision system for real-time ball tracking and spin estimation, combining an event camera with a focus-tunable telephoto lens and fast galvanometer mirrors to obtain a magnified, motion-compensated view of unmodified balls in flight. The hybrid tracking framework fuses 3D ball localization with 2D event-based detection for smooth, robust active tracking at professional ball-sport speeds. The offline spin estimation method,…

Figures reproduced from arXiv:2606.26780 by the authors.

References

How the literature uses it

What this paper does with the arXiv papers it cites

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

What it cites · 4 sentences without a confident label

Uses the method or tool of 1 paper

Discussion

Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.