Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

ETAP: Event-based Tracking of Any Point

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read An event-only tracker, trained on synthetic events, outperforms prior event-only and events-plus-frames methods and tracks points where frame-based trackers fail.

desk verdict First event-only TAP that genuinely works across real datasets; the E2D2 ground truth is the one place to press, but the core claims hold. read the letter →

arxiv 2412.00133 v2 pith:SCC5245M submitted 2024-11-28 cs.CV cs.LGcs.RO

Friedhelm Hamann , Daniel Gehrig , Filbert Febryanto , Kostas Daniilidis , Guillermo Gallego This is my paper · ORCID
classification cs.CVcs.LGcs.RO
keywords eventcamerastrackinganypointevent-basedvisionsyntheticdatacontrastivefeaturealignmentmotionrobustness
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

The paper sets out to show that event cameras, which output asynchronous brightness-change spikes, can serve as the sole input for tracking arbitrary points in a video, a task previously dominated by frame-based tracking-any-point methods. It introduces ETAP, an event-only TAP model built by adapting a transformer point-tracker to event-stack representations, and trains it on EventKubric, a new synthetic dataset rendered at high frame rate and converted to events. The central result is that this event-only tracker outperforms the previous best event-only feature tracker by about 20% (feature age 0.704 vs 0.589), surpasses the best events-plus-frames TAP method by 4.1%, and improves average Jaccard by 136% over an event-to-frame baseline on the TAP benchmark. If correct, this matters because event cameras handle high-speed motion, motion blur, and high dynamic range where frame-based trackers fail, making reliable point tracks available in exactly the regimes that break conventional video. The paper also claims a new contrastive feature-alignment loss that makes event features invariant to motion direction, plus new ground truth for two real datasets.

What carries the argument

The load-bearing object is the event stack: each tracking timestep is represented by a 10-channel histogram of the most recent $N_e$ events, binned at hierarchical time resolutions, so asynchronous events become grid tensors a convolutional encoder and transformer can consume. Around this, the tracker uses the same iterative token-refinement mechanism as modern frame-based point trackers: tokens carry position offsets, per-point descriptors, spatial correlation features (inner products of a descriptor against a 7×7 neighborhood at four scales), and visibility flags, updated by alternating intra-point and temporal attention. The specifically new mechanism is the feature-alignment loss $\mathcal{L}_{fa}$: for each training sample a time-inverted and randomly rotated duplicate is generated, and descriptors sampled from real and inverted tracks are matched after reversing time, rewarding cosine similarity so that learned features depend on scene structure rather than on the direction of motion. The third piece is the training data: EventKubric, generated by rendering 2-second physically based videos at 48 fps, adaptively upsampling so inter-frame motion is at most one pixel, and converting to events with a simulator under randomized contrast thresholds.

What would settle it

Rerun the E2D2 fidget-spinner experiment with independent ground truth, such as a high-speed camera, motion capture, or a checkerboard-encoded marker, and with a non-piecewise-constant velocity model; if ETAP's average-Jaccard margin over frame-based and event-to-frame baselines collapses or reverses when the assumed constant-velocity ground truth is replaced, the central claim fails. A second, easier check is to vary the valley-detection L2-norm threshold (300), fit a smooth angular-velocity model, and recompute the reported 0.308 average Jaccard.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that an event-only neural tracker can track arbitrary points, not just corners, robustly enough to beat both prior event-only methods and methods that combine events with frames. The model takes query points, builds multi-channel event stacks from the latest fixed number of events, extracts multi-scale feature maps, and refines point positions and appearance descriptors through a transformer with spatial correlations and visibility prediction, in the style of modern frame-based TAP trackers. Trained purely on synthetic EventKubric clips, it reports feature age 0.704 on EDS versus 0.589 for the best prior event-only method and 0.676 for the best events-plus-frames method; on the TAP benchmark it reaches average Jaccard 0.546 on EventKubric and 0.661 on EVIMO2, versus 0.236 and 0.531 for an event-to-frame reconstruction baseline. The second claimed discovery is that the motion-dependence of event features can be tamed by a feature-alignment loss: each training sample is duplicated under time inversion plus random rotation, and descriptors at corresponding track points are pulled together in cosine similarity, shrinking the inter-motion similarity gap from 0.377 to 0.067 in their measurement.

Load-bearing premise

The method's advantage over frame-based trackers in its headline high-speed demonstration is measured against self-generated ground truth for the E2D2 fidget spinner that assumes constant angular velocity between valleys detected by a hand-set threshold, even though the spinner accelerates; if that assumption is wrong, the quantified advantage changes.

Editorial extensions

If this is right

  • Event-only tracking can be deployed in high-speed and high-dynamic-range settings where frame-based trackers fail, including low-light recordings with severe motion blur.
  • The 20% feature-age gain over the best prior event-only method and 4.1% over the best events-plus-frames method suggest events alone carry enough information for long-range point correspondence.
  • The feature-alignment loss makes learned event features stable under opposite and perpendicular motion directions, closing most of the similarity gap between same-motion and different-motion descriptors.
  • The EventKubric pipeline, with high render frame rate and randomized contrast thresholds, transfers to real sensors of different resolutions and camera types without fine-tuning.
  • New ground truth tracks provided for EVIMO2 and E2D2 allow future event-based TAP methods to be compared directly on real high-speed data.

Reading between the lines

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

  • The feature-alignment idea is not limited to point tracks: the same time-inversion and rotation invariance principle could be applied to event representations for optical flow, segmentation, or recognition, wherever descriptor drift from motion direction is a problem.
  • Because the paper reports that query times during absence of motion produce poor descriptors, an event-only system in practice may need a motion-gating or reinitialization heuristic; a testable extension is to couple ETAP with an inertial or frame-based wake-up to handle static starts.
  • If event-only TAP sustains its margins on more diverse real data, downstream systems like visual odometry or robot manipulation could drop the frame stream entirely at high speed, saving bandwidth and power; this is a consequence the paper does not develop.
  • A direct stress test would be evaluating on slow, low-texture scenes where event data is sparse; the paper's limitation note suggests accuracy may degrade, and quantifying that would set the boundary of the method.
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

3 major / 5 minor

Summary. This paper introduces ETAP, an event-only tracking-any-point (TAP) model. The method processes event stacks with a CoTracker-style transformer and is trained on a new synthetic EventKubric dataset produced by combining Kubric rendering with Vid2e/ESIM event generation. A time-inversion contrastive feature-alignment loss is proposed to make descriptors robust to motion direction. Experiments cover TAP on EventKubric, EVIMO2, and E2D2, and feature tracking on EDS and EC, reporting improvements over event-only and frame-event fusion baselines. Code and dataset are released.

Significance. If the reported results hold, the paper makes a strong contribution: it is the first event-only TAP approach, with released code and dataset, a systematic ablation study, and a dedicated experiment showing that the feature-alignment loss improves motion-invariance of descriptors. Cross-dataset generalization to different sensors and resolutions is demonstrated. The main caveat is the self-generated E2D2 ground truth, which directly supports the high-speed/HDR claim; this needs additional validation.

major comments (3)
  1. [Supplementary §8.1, Table 3] The E2D2 fidget-spinner ground truth is generated from event histograms using a hand-set L2-norm valley threshold of 300 and a piecewise-constant angular-velocity model between detected third-revolution valleys, yet the text explicitly states that the spinner accelerates throughout the clip. Because both ETAP and the baselines are scored against this synthetic GT, the reported AJ gap (0.308 vs 0.183) is not independently validated, and the qualitative claim that frame-based tracking fails while ETAP succeeds rests on this assumption. Please provide a sensitivity analysis over the threshold and an alternative acceleration-aware velocity model, or an independent manual annotation check, and temper the claims if the quantitative advantage is not robust.
  2. [Abstract; §5.2, Table 2] The abstract's '136% better' figure corresponds to ETAP without the FA-loss (0.556 vs 0.236 on EventKubric, a 135.6% improvement) rather than the final ETAP model (0.546 vs 0.236, a 131.4% improvement). Since the final model is the one presented as the method, please correct the headline number or explicitly state that the improvement is computed for the ablation without the feature-alignment loss.
  3. [§5.2, Table 2] The 'E2Vid [52] + CoTracker [29]' baseline is an events-to-video pipeline rather than a pure event-only method; the text calls it an event-based baseline and the distinction matters because events are first converted to frames. Please clarify this in the main text and in the table caption so that readers can correctly interpret the comparison.
minor comments (5)
  1. [Table 5] The 'Varying dynamics' and 'static' configurations are not defined; please state whether 'static' means no camera motion, no object motion, or both, and explain how panning is incorporated.
  2. [Tables 2-4] The notation 'w\o' should be 'w/o' for readability.
  3. [Figure 10] Figure 10(b) truncates the L2-norm axis at 400, making the detected valleys difficult to inspect; a zoomed view around the threshold region would help the reader judge the reliability of the valley detection.
  4. [Supplementary §7] The time-inversion derivation relies on the approximation δτ~ ≈ δτ; please state this approximation explicitly when the feature-alignment loss is motivated in the main text.
  5. [§5.2] The statement that ETAP surpasses the event-based baseline by '68% AJ' should specify that this is a relative improvement (0.308 vs 0.183), to avoid confusion with an absolute difference of 0.125.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: ETAP's architecture, losses, and benchmark results are not equivalent to their inputs; E2D2 GT concerns are evaluation validity, not circularity.

full rationale

The paper's derivation chain is empirical and self-contained. The tracker follows the external CoTracker architecture [29] rather than a self-cited uniqueness theorem; training uses the external Kubric/Vid2e/ESIM pipeline; and the two headline experimental claims are measured against external benchmarks (EDS/EC feature tracking, EventKubric held-out test split) with independent baselines (E2VID+CoTracker, FE-TAP). The feature-alignment loss is derived from the linearized event generation model with a stated approximation (δτ-tilde ≈ δτ) and is used as a training regularizer, not as a source of the reported numbers. No fitted parameter is renamed as a prediction: the E2D2 ground truth is hand-generated from a thresholded event-histogram valley detector and a piecewise-constant angular-velocity model, and both ETAP and the baselines are scored against that same GT, so any weakness there concerns the validity/accuracy of the evaluation, not a circular reduction of the method's output to its input. The only self-citations (E2D2 [60], EVIMO2 methodology [24]) support data provenance and evaluation protocol, not the load-bearing derivation, and the Fidget-Spinner GT assumption is explicitly disclosed in Supplementary Sec. 8.1. The Limitations section (Sec. 5.5) also openly acknowledges a genuine weakness (query-time dependence) rather than hiding it, further supporting a non-circular reading.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard deep-learning practice plus four domain assumptions: transfer of the CoTracker architecture to event stacks, LEGM-based time-inversion reasoning with an explicit approximation, synthetic-to-real fidelity of EventKubric, and a geometric model for the E2D2 ground truth. Free parameters are conventional hyperparameters plus the hand-set E2D2 valley threshold. No new physical entities are postulated; the time-inverted training sample is an augmentation, not an entity.

free parameters (5)
  • contrast sensitivity range = C ~ U(0.16, 0.34)
    Used for all EventKubric event generation (Section 4) and selected based on reference [30]; ablations tested fixed 0.2 and U(0.20, 1.50), with U(0.16, 0.34) best on EDS (Table 5).
  • events per stack Ne = 4 x 10^5
    Number of events aggregated into each event-stack representation (Section 5.1, Table 6); chosen by hand without a reported sensitivity study.
  • loss weights = 0.1*Ltp + Lvis + 0.1*Lfa
    Total loss weighting defined in Section 3; weights chosen by hand, no ablation of the 0.1 coefficients is reported.
  • tracker hyperparameters = w=8, stride Ts=4, bins B=10, patch radius delta=3, refinement M=4 train / 6 eval
    Window length, stride, bin number, correlation patch size, and refinement iterations (Table 6); carried over from CoTracker-style design with minor adjustment.
  • E2D2 ground-truth valley threshold = 300 (L2 norm)
    Hand-set threshold for detecting third-revolution valleys in the event-histogram time series used to build the E2D2 fidget-spinner ground truth (Supplementary Section 8.1); this threshold defines the GT timestamps.
assumptions (4)
  • domain assumption CoTracker's transformer tracker and its inductive biases transfer to event-stack input features.
    The update rule in equations (3) to (5) and the intra-point and temporal attention design follow reference [29] (Section 3, Tracker); the paper assumes these biases remain effective when features are extracted from event stacks instead of frames.
  • domain assumption Linearized event generation model (LEGM) with delta-tau-tilde approximately equal to delta-tau under time inversion.
    The feature-alignment loss motivation derives that inverted events differ from originals using LEGM and the explicit approximation that the time since the last event is similar under time inversion (Supplementary Section 7, equations 8 to 9); the empirical ablation in Table 7 provides the actual support.
  • domain assumption Events generated by Kubric + FILM + ESIM are representative enough of real event camera data for cross-dataset transfer.
    Training on EventKubric underlies all real-dataset results (Section 4); strong cross-dataset numbers on EDS, EC, EVIMO2, and E2D2 give indirect support, but sensor noise, HDR behavior, and non-idealities are not explicitly modeled.
  • domain assumption The E2D2 fidget spinner wheel is a perfect circle facing the camera, with angular velocity constant between detected third-revolution valleys.
    Ground-truth tracks for the E2D2 evaluation are derived geometrically from this model (Supplementary Section 8.1); violations directly affect the reported advantage over baselines in Table 3.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ETAP: Event-based Tracking of Any Point." pith.science (2026). https://pith.science/paper/SCC5245M

@misc{pith2026241200133,
  author       = {Pith},
  title        = {Pith review of: ETAP: Event-based Tracking of Any Point},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SCC5245M}},
  note         = {Machine review of arXiv:2412.00133}
}
read the original abstract

Tracking any point (TAP) recently shifted the motion estimation paradigm from focusing on individual salient points with local templates to tracking arbitrary points with global image contexts. However, while research has mostly focused on driving the accuracy of models in nominal settings, addressing scenarios with difficult lighting conditions and high-speed motions remains out of reach due to the limitations of the sensor. This work addresses this challenge with the first event camera-based TAP method. It leverages the high temporal resolution and high dynamic range of event cameras for robust high-speed tracking, and the global contexts in TAP methods to handle asynchronous and sparse event measurements. We further extend the TAP framework to handle event feature variations induced by motion -- thereby addressing an open challenge in purely event-based tracking -- with a novel feature-alignment loss which ensures the learning of motion-robust features. Our method is trained with data from a new data generation pipeline and systematically ablated across all design decisions. Our method shows strong cross-dataset generalization and performs 136% better on the average Jaccard metric than the baselines. Moreover, on an established feature tracking benchmark, it achieves a 20% improvement over the previous best event-only method and even surpasses the previous best events-and-frames method by 4.1%. Our code is available at https://github.com/tub-rip/ETAP

Figures

Figures reproduced from arXiv: 2412.00133 by the authors.

Figure 1
Figure 1. Event-only Tracking of Any Point. Our method uses only events to track semi-dense long-range point trajectories, working in conditions where frame-based methods fail. (HDR), and low power consumption, characteristics that make them valuable stand-alone sensors for various robotic perception tasks. These innovative sensors address several limitations of conventional cameras, particularly handling motion blur and high… view at source ↗
Figure 2
Figure 2. The motion dependence problem. Many tracking meth￾ods rely on the correspondence of features. While the appear￾ance of frames (left) is independent of the scene movement, the event camera data depends on the motion direction. Image cour￾tesy of [1]. event-based methods. TAP is evaluated on EventKubric, the Extreme Event Decompression Dataset (E2D2) [60] (for which we provide new ground truth), and on custom se￾quenc… view at source ↗
Figure 3
Figure 3. (a) During training each sample has a time inverted duplicate, model [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Task 1 - TAP on EVIMO2. Visualization of track predic￾tions from first to last timestamp. Queries 𝑞𝑖 = 𝑥𝑡𝑖 𝑖 Ours GT [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Task 1 - TAP on EventKubric. Semi-dense tracks are predicted for 2s-samples. Metrics Method Input AJ ↑ δ x avg ↑ OA ↑ EVIMO2 E2Vid [52] + CoTracker [29] Events 0.531 0.663 0.861 ETAP w\o FA-loss (Ours) Events 0.655 0.787 0.884 ETAP (Ours) Events 0.661 0.789 0.895 Event…
Figure 6
Figure 6. Figure 6: Task 1 - TAP on E2D2. Shown are four timesteps of each sequence. At the beginning (t0) the model is queried with the marked points. Input modality: F - frames, E - Events tracks from its motion capture data. Our approach mirrors the EVIMO2 Continuous Flow Dataset [24] …
Figure 8
Figure 8. Figure 8: Task 2 - Feature tracking on EDS. Notably, our tracker captures points even after they leave the FOV and reenter. EDS EC Method Input Feature Age ↑ Expected FA ↑ Feature Age ↑ Expected FA ↑ ICP [32] E 0.060 0.040 0.256 0.245 EKLT [21] E+F 0.325 0.205 0.811 0.775 DDFT […
Figure 7
Figure 7. Figure 7: Task 1 - TAP qualitative result. Tracking in a very de￾manding scenario: a small, low-textured, fast object with high de￾formation and an HDR background. cross-modality comparison, specifically picking a scene of a rotating fidget spinner ( [PITH_FULL_IMAGE:figures/fu…
Figure 9
Figure 9. Figure 9: Asynchronous events are converted into temporally [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 11
Figure 11. Figure 11: visualizes the data generation explained in Sec. 4 [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]
Figure 13
Figure 13. Figure 13: A few examples of EventKubric. Point tracks are subsampled for better visualization. [PITH_FULL_IMAGE:figures/full_fig_p012_13.png]
Figure 14
Figure 14. Figure 14: Task 1 - TAP on EVIMO2 data. Visualization of track predictions. 13 [PITH_FULL_IMAGE:figures/full_fig_p013_14.png]
Figure 15
Figure 15. Figure 15: Additional visualizations on the EDS and EC dataset. [PITH_FULL_IMAGE:figures/full_fig_p014_15.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. MATE: Motion-Augmented Temporal Consistency for Event-based Point Tracking

    cs.CV 2024-12 conditional novelty 6.0 of 10

    MATE tracks any point from event cameras alone, using motion vectors extracted from time surfaces to guide matching, and reports higher accuracy and survival than video- and event-based baselines on four benchmarks.

  2. Static in Frames, Dynamic in Events: Rethinking Features in Event Cameras as Motion Cues

    cs.CV 2026-08 conditional novelty 5.0 of 10

    Harris eigenvalues and spatiotemporal density values from event cameras encode motion direction and, when added to an optical flow network, improve accuracy in data-scarce settings.

Reference graph

Works this paper leans on

64 extracted references · 60 canonical work pages · cited by 2 Pith papers

  1. [52]

    Events-to-video: Bringing modern computer vision to event cameras

    Henri Rebecq, Ren ´e Ranftl, Vladlen Koltun, and Davide Scaramuzza. Events-to-video: Bringing modern computer vision to event cameras. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR), 2019. 5, 6

    work page 2019

  2. [29]

    Co- Tracker: It is better to track together

    Nikita Karaev, Ignacio Rocco, Benjamin Graham, Natalia Neverova, Andrea Vedaldi, and Christian Rupprecht. Co- Tracker: It is better to track together. Eur. Conf. Comput. Vis. (ECCV), 2024. 3, 4, 5, 6, 7

    work page 2024

  3. [1]

    Event-driven Feature Detection and Tracking for Visual SLAM

    Ignacio Alzugaray. Event-driven Feature Detection and Tracking for Visual SLAM. PhD thesis, ETH Zurich, 2022. 2

    work page 2022

  4. [2]

    ACE: An efficient asynchronous corner tracker for event cameras

    Ignacio Alzugaray and Margarita Chli. ACE: An efficient asynchronous corner tracker for event cameras. In Int. Conf. 3D Vision (3DV), pages 653–661, 2018. 3

    work page 2018

  5. [3]

    Haste: multi- hypothesis asynchronous speeded-up tracking of events

    Ignacio Alzugaray and Margarita Chli. Haste: multi- hypothesis asynchronous speeded-up tracking of events. In British Mach. Vis. Conf. (BMVC), page 744, 2020. 7, 11

    work page 2020

  6. [4]

    Lucas-kanade 20 years on: A unifying frame- work: Part 2

    Simon Baker, Ralph Gross, Ishikawa Takahiro, and Iain Matthews. Lucas-kanade 20 years on: A unifying frame- work: Part 2. Technical Report CMU-RI-TR-03-01, 2003. 2

    work page 2003

  7. [5]

    Blender—a 3d modelling and rendering package

    D Blender Online Community. Blender—a 3d modelling and rendering package. Blender Foundation, 2018. 5

    work page 2018

  8. [6]

    EVIMO2: An event camera dataset for motion segmentation, optical flow, structure from motion, and visual inertial odometry in indoor scenes with monoc- ular or stereo algorithms

    Levi Burner, Anton Mitrokhin, Cornelia Ferm ¨uller, and Yiannis Aloimonos. EVIMO2: An event camera dataset for motion segmentation, optical flow, structure from motion, and visual inertial odometry in indoor scenes with monoc- ular or stereo algorithms. arXiv e-prints, 2022. 6

    work page 2022

Show all 64 references
  1. [7]

    LEAP-VO: Long-term effective any point tracking for vi- sual odometry

    Weirong Chen, Le Chen, Rui Wang, and Marc Pollefeys. LEAP-VO: Long-term effective any point tracking for vi- sual odometry. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR), pages 19844–19853, 2024. 1

    2024

  2. [8]

    TallFormer: Temporal action localization with a long-memory transformer

    Feng Cheng and Gedas Bertasius. TallFormer: Temporal action localization with a long-memory transformer. In Eur. Conf. Comput. Vis. (ECCV), pages 503–521, 2022. 3

    2022

  3. [9]

    Detecting stable keypoints from events through im- age gradient prediction

    Philippe Chiberre, Etienne Perot, Amos Sironi, and Vincent Lepetit. Detecting stable keypoints from events through im- age gradient prediction. In IEEE Conf. Comput. Vis. Pattern Recog. Workshops (CVPRW), 2021. 3

    2021

  4. [10]

    Local all-pair correspon- dence for point tracking

    Seokju Cho, Jiahui Huang, Jisu Nam, Honggyu An, Seun- gryong Kim, and Joon-Young Lee. Local all-pair correspon- dence for point tracking. Eur. Conf. Comput. Vis. (ECCV),

  5. [11]

    Bullet physics simulation

    Erwin Coumans. Bullet physics simulation. In ACM SIG- GRAPH 2015 Courses, 2015. 5

    2015

  6. [12]

    A voxel graph cnn for object classification with event cameras

    Yongjian Deng, Hao Chen, Hai Liu, and Youfu Li. A voxel graph cnn for object classification with event cameras. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR) , pages 1172–1181, 2022. 4

    2022

  7. [13]

    TAP-Vid: A benchmark for track- ing any point in a video

    Carl Doersch, Ankush Gupta, Larisa Markeeva, Adria Re- casens, Lucas Smaira, Yusuf Aytar, Joao Carreira, Andrew Zisserman, and Yi Yang. TAP-Vid: A benchmark for track- ing any point in a video. In Adv. Neural Inf. Process. Syst. (NeurIPS), pages 13610–13626, 2022. 1, 3, 6

    2022

  8. [14]

    Tapir: Tracking any point with per-frame initialization and temporal refinement

    Carl Doersch, Yi Yang, Mel Vecerik, Dilara Gokay, Ankush Gupta, Yusuf Aytar, Joao Carreira, and Andrew Zisserman. Tapir: Tracking any point with per-frame initialization and temporal refinement. In Int. Conf. Comput. Vis. (ICCV) , pages 10061–10072, 2023. 3, 6

    2023

  9. [15]

    FlowNet: Learn- ing optical flow with convolutional networks

    Alexey Dosovitskiy, Philipp Fischer, Eddy Ilg, Philip H¨ausser, Caner Hazırbas ¸, Vladimir Golkov, Patrick van der Smagt, Daniel Cremers, and Thomas Brox. FlowNet: Learn- ing optical flow with convolutional networks. In Int. Conf. Comput. Vis. (ICCV), pages 2758–2766, 2015. 1, 3

    2015

  10. [16]

    Qdtrack: Quasi-dense similarity learning for appearance-only multi- ple object tracking

    Tobias Fischer, Jiangmiao Pang, Thomas E Huang, Linlu Qiu, Haofeng Chen, Trevor Darrell, and Fisher Yu. Qdtrack: Quasi-dense similarity learning for appearance-only multi- ple object tracking. arXiv preprint arXiv:2210.06984, 2022. 3

    arXiv

  11. [17]

    Event-based vision: A survey

    Guillermo Gallego, Tobi Delbruck, Garrick Orchard, Chiara Bartolozzi, Brian Taba, Andrea Censi, Stefan Leutenegger, Andrew Davison, J ¨org Conradt, Kostas Daniilidis, and Da- vide Scaramuzza. Event-based vision: A survey. IEEE Trans. Pattern Anal. Mach. Intell., 44(1):154–180,...

    2022

  12. [18]

    Asynchronous, photometric feature track- ing using events and frames

    Daniel Gehrig, Henri Rebecq, Guillermo Gallego, and Da- vide Scaramuzza. Asynchronous, photometric feature track- ing using events and frames. In Eur. Conf. Comput. Vis. (ECCV), pages 766–781, 2018. 3

    2018

  13. [19]

    Derpa- nis, and Davide Scaramuzza

    Daniel Gehrig, Antonio Loquercio, Konstantinos G. Derpa- nis, and Davide Scaramuzza. End-to-end learning of repre- sentations for asynchronous event-based data. In Int. Conf. Comput. Vis. (ICCV), pages 5632–5642, 2019. 4

    2019

  14. [20]

    Video to Events: Recycling video datasets for event cameras

    Daniel Gehrig, Mathias Gehrig, Javier Hidalgo-Carri ´o, and Davide Scaramuzza. Video to Events: Recycling video datasets for event cameras. In IEEE Conf. Comput. Vis. Pat- tern Recog. (CVPR), pages 3583–3592, 2020. 2, 6

    2020

  15. [21]

    EKLT: Asynchronous photometric feature tracking using events and frames

    Daniel Gehrig, Henri Rebecq, Guillermo Gallego, and Da- vide Scaramuzza. EKLT: Asynchronous photometric feature tracking using events and frames. Int. J. Comput. Vis., 128: 601–618, 2020. 3, 7, 11

    2020

  16. [22]

    Dense continuous-time optical flow from event cameras

    Mathias Gehrig, Manasi Muglikar, and Davide Scaramuzza. Dense continuous-time optical flow from event cameras. IEEE Trans. Pattern Anal. Mach. Intell. , 46(7):4736–4746,

  17. [23]

    Kubric: A scalable dataset generator

    Klaus Greff, Francois Belletti, Lucas Beyer, Carl Doersch, Yilun Du, Daniel Duckworth, David J Fleet, Dan Gnanapra- gasam, Florian Golemo, Charles Herrmann, et al. Kubric: A scalable dataset generator. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR), pages 3749–3761, 2022. 2, 3, 5

    2022

  18. [24]

    Motion- prior contrast maximization for dense continuous-time mo- tion estimation

    Friedhelm Hamann, Ziyun Wang, Ioannis Asmanis, Kenneth Chaney, Guillermo Gallego, and Kostas Daniilidis. Motion- prior contrast maximization for dense continuous-time mo- tion estimation. In Eur. Conf. Comput. Vis. (ECCV), 2024. 3, 6

    2024

  19. [25]

    Particle video revisited: Tracking through occlusions using point trajectories

    Adam W Harley, Zhaoyuan Fang, and Katerina Fragkiadaki. Particle video revisited: Tracking through occlusions using point trajectories. In Eur. Conf. Comput. Vis. (ECCV), pages 59–75, 2022. 1, 3

    2022

  20. [26]

    Event-aided direct sparse odometry

    Javier Hidalgo-Carri ´o, Guillermo Gallego, and Davide Scaramuzza. Event-aided direct sparse odometry. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR) , pages 5781– 5790, 2022. 2, 7

    2022

  21. [27]

    Horn and Brian G

    Berthold K.P. Horn and Brian G. Schunck. Determining op- tical flow. J. Artificial Intell., 17(1):185 – 203, 1981. 3

    1981

  22. [28]

    FlowNet 2.0: Evolution of optical flow estimation with deep networks

    Eddy Ilg, Nikolaus Mayer, Tonmoy Saikia, Margret Keu- per, Alexey Dosovitskiy, and Thomas Brox. FlowNet 2.0: Evolution of optical flow estimation with deep networks. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR) , pages 1647–1655, 2017. 3 15

    2017

  23. [30]

    Deep event visual odometry

    Simon Klenk, Marvin Motzet, Lukas Koestler, and Daniel Cremers. Deep event visual odometry. In Int. Conf. 3D Vi- sion (3DV), pages 739–749, 2024. 6, 8

    2024

  24. [31]

    The ninth visual object tracking vot2021 challenge results

    Matej Kristan, Ji ˇr´ı Matas, Aleˇs Leonardis, Michael Felsberg, Roman Pflugfelder, Joni-Kristian K ¨am¨ar¨ainen, Hyung Jin Chang, Martin Danelljan, Luka Cehovin, Alan Lukeˇziˇc, et al. The ninth visual object tracking vot2021 challenge results. In Int. Conf. Comput. Vis. (ICC...

    2021

  25. [32]

    Low-latency visual odometry using event- based feature tracks

    Beat Kueng, Elias Mueggler, Guillermo Gallego, and Da- vide Scaramuzza. Low-latency visual odometry using event- based feature tracks. In IEEE/RSJ Int. Conf. Intell. Robot. Syst. (IROS), pages 16–23, 2016. 3, 7, 11

    2016

  26. [33]

    Asynchronous event-based multiker- nel algorithm for high-speed visual features tracking

    Xavier Lagorce, C ´edric Meyer, Sio-Hoi Ieng, David Filliat, and Ryad Benosman. Asynchronous event-based multiker- nel algorithm for high-speed visual features tracking. IEEE Trans. Neural Netw. Learn. Syst., 26(8):1710–1720, 2015. 3

    2015

  27. [34]

    A survey of appear- ance models in visual object tracking

    Xi Li, Weiming Hu, Chunhua Shen, Zhongfei Zhang, An- thony Dick, and Anton Van Den Hengel. A survey of appear- ance models in visual object tracking. ACM transactions on Intelligent Systems and Technology (TIST), 4(4):1–48, 2013. 2

    2013

  28. [35]

    Blinkflow: A dataset to push the limits of event-based op- tical flow estimation

    Yijin Li, Zhaoyang Huang, Shuo Chen, Xiaoyu Shi, Hong- sheng Li, Hujun Bao, Zhaopeng Cui, and Guofeng Zhang. Blinkflow: A dataset to push the limits of event-based op- tical flow estimation. In IEEE/RSJ Int. Conf. Intell. Robot. Syst. (IROS), pages 3881–3888, 2023. 3

    2023

  29. [36]

    A 128×128 120 dB 15 µs latency asynchronous temporal con- trast vision sensor

    Patrick Lichtsteiner, Christoph Posch, and Tobi Delbruck. A 128×128 120 dB 15 µs latency asynchronous temporal con- trast vision sensor. IEEE J. Solid-State Circuits, 43(2):566– 576, 2008. 3

    2008

  30. [37]

    Bauer, Ahmed Nabil Belbachir, P

    Martin Litzenberger, Christoph Posch, D. Bauer, Ahmed Nabil Belbachir, P. Sch ¨on, B. Kohn, and H. Garn. Embedded vision system for real-time object tracking using an asynchronous transient vision sensor. In Digital Signal Processing Workshop, pages 173–178, 2006. 3

    2006

  31. [38]

    Tracking any point with frame- event fusion network at high frame rate

    Jiaxiong Liu, Bo Wang, Zhen Tan, Jinpu Zhang, Hui Shen, and Dewen Hu. Tracking any point with frame- event fusion network at high frame rate. arXiv preprint arXiv:2409.11953, 2024. 3, 7, 11

    2024 arXiv

  32. [39]

    Decoupled weight decay regularization

    I Loshchilov. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017. 6

    2017 arXiv

  33. [40]

    Lucas and Takeo Kanade

    Bruce D. Lucas and Takeo Kanade. An iterative image regis- tration technique with an application to stereo vision. In Int. Joint Conf. Artificial Intell. (IJCAI), pages 674–679, 1981. 2

    1981

  34. [41]

    Speed invariant time surface for learning to detect corner points with event- based cameras

    Jacques Manderscheid, Amos Sironi, Nicolas Bourdis, Da- vide Migliore, and Vincent Lepetit. Speed invariant time surface for learning to detect corner points with event- based cameras. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR), 2019. 3

    2019

  35. [42]

    The template update problem

    Iain Matthews, Takahiro Ishikawa, and Simon Baker. The template update problem. IEEE Trans. Pattern Anal. Mach. Intell., 2004. 2

    2004

  36. [43]

    A large dataset to train convolutional networks for disparity, optical flow, and scene flow estimation

    Nikolaus Mayer, Eddy Ilg, Philip Hausser, Philipp Fischer, Daniel Cremers, Alexey Dosovitskiy, and Thomas Brox. A large dataset to train convolutional networks for disparity, optical flow, and scene flow estimation. In IEEE Conf. Com- put. Vis. Pattern Recog. (CVPR), pages 404...

    2016

  37. [44]

    Data-driven feature tracking for event cameras

    Nico Messikommer, Carter Fang, Mathias Gehrig, and Da- vide Scaramuzza. Data-driven feature tracking for event cameras. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR), 2023. 1, 3, 7, 11, 14

    2023

  38. [45]

    The event-camera dataset and simulator: Event-based data for pose estimation, visual odometry, and SLAM

    Elias Mueggler, Henri Rebecq, Guillermo Gallego, Tobi Del- bruck, and Davide Scaramuzza. The event-camera dataset and simulator: Event-based data for pose estimation, visual odometry, and SLAM. Int. J. Robot. Research , 36(2):142– 149, 2017. 7

    2017

  39. [46]

    Stereo depth from events cameras: Concen- trate and focus on the future

    Yeongwoo Nam, Mohammad Mostafavi, Kuk-Jin Yoon, and Jonghyun Choi. Stereo depth from events cameras: Concen- trate and focus on the future. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR), pages 6104–6113, 2022. 4, 9

    2022

  40. [47]

    Visual tracking using neu- romorphic asynchronous event-based cameras

    Zhenjiang Ni, Sio-Ho ¨ı Ieng, Christoph Posch, St ´ephane R´egnier, and Ryad Benosman. Visual tracking using neu- romorphic asynchronous event-based cameras. Neural Com- putation, 27(4):925–953, 2015. 3

    2015

  41. [48]

    Quasi-dense similarity learn- ing for multiple object tracking

    Jiangmiao Pang, Linlu Qiu, Xia Li, Haofeng Chen, Qi Li, Trevor Darrell, and Fisher Yu. Quasi-dense similarity learn- ing for multiple object tracking. In IEEE Conf. Comput. Vis. Pattern Recog. (CVPR), 2021. 3

    2021

  42. [49]

    The 2017 DA VIS challenge on video object segmentation.arXiv preprint arXiv:1704.00675, 2017

    Jordi Pont-Tuset, Federico Perazzi, Sergi Caelles, Pablo Ar- bel´aez, Alex Sorkine-Hornung, and Luc Van Gool. The 2017 DA VIS challenge on video object segmentation.arXiv preprint arXiv:1704.00675, 2017. 3

    2017 arXiv

  43. [50]

    An asynchronous time-based image sensor

    Christoph Posch, Daniel Matolin, and Rainer Wohlgenannt. An asynchronous time-based image sensor. In IEEE Int. Symp. Circuits Syst. (ISCAS), pages 2130–2133, 2008. 3

    2008

  44. [51]

    ESIM: an open event camera simulator

    Henri Rebecq, Daniel Gehrig, and Davide Scaramuzza. ESIM: an open event camera simulator. InConf. on Robotics Learning (CoRL), pages 969–982. PMLR, 2018. 5, 6

    2018

  45. [53]

    FILM: Frame in- terpolation for large motion

    Fitsum Reda, Janne Kontkanen, Eric Tabellion, Deqing Sun, Caroline Pantofaru, and Brian Curless. FILM: Frame in- terpolation for large motion. In Eur. Conf. Comput. Vis. (ECCV), pages 250–266, 2022. 5

    2022

  46. [54]

    Particle video: Long-range mo- tion estimation using point trajectories

    Peter Sand and Seth Teller. Particle video: Long-range mo- tion estimation using point trajectories. Int. J. Comput. Vis., 80:72–91, 2008. 1, 3

    2008

  47. [55]

    Event-based background oriented schlieren

    Shintaro Shiba, Friedhelm Hamann, Yoshimitsu Aoki, and Guillermo Gallego. Event-based background oriented schlieren. IEEE Trans. Pattern Anal. Mach. Intell. , 46(4): 2011–2026, 2024. 3

    2011

  48. [56]

    Reducing the sim-to-real gap for event cam- eras

    Timo Stoffregen, Cedric Scheerlinck, Davide Scaramuzza, Tom Drummond, Nick Barnes, Lindsay Kleeman, and Robert Mahony. Reducing the sim-to-real gap for event cam- eras. In Eur. Conf. Comput. Vis. (ECCV) , pages 534–549,

  49. [57]

    Feature detection and tracking with the dy- namic and active-pixel vision sensor (DA VIS)

    David Tedaldi, Guillermo Gallego, Elias Mueggler, and Da- vide Scaramuzza. Feature detection and tracking with the dy- namic and active-pixel vision sensor (DA VIS). In Int. Conf. 16 Event-Based Control, Comm. Signal Proc. (EBCCSP), 2016. 3

    2016

  50. [58]

    RAFT: Recurrent all pairs field transforms for optical flow

    Zachary Teed and Jia Deng. RAFT: Recurrent all pairs field transforms for optical flow. In Eur. Conf. Comput. Vis. (ECCV), pages 402–419, 2020. 1, 3

    2020

  51. [59]

    Robotap: Tracking arbitrary points for few-shot visual imitation

    Mel Vecerik, Carl Doersch, Yi Yang, Todor Davchev, Yusuf Aytar, Guangyao Zhou, Raia Hadsell, Lourdes Agapito, and Jon Scholz. Robotap: Tracking arbitrary points for few-shot visual imitation. In IEEE Int. Conf. Robot. Autom. (ICRA) , pages 5397–5403, 2024. 1

    2024

  52. [60]

    Event- based continuous color video decompression from single frames

    Ziyun Wang, Friedhelm Hamann, Kenneth Chaney, Wen Jiang, Guillermo Gallego, and Kostas Daniilidis. Event- based continuous color video decompression from single frames. arXiv preprint arXiv:2312.00113, 2023. 2, 6

    2023 arXiv

  53. [61]

    Scalable video object segmentation with simplified frame- work

    Qiangqiang Wu, Tianyu Yang, Wei Wu, and Antoni B Chan. Scalable video object segmentation with simplified frame- work. In Int. Conf. Comput. Vis. (ICCV) , pages 13879– 13889, 2023. 3

    2023

  54. [62]

    Pointodyssey: A large-scale synthetic dataset for long-term point tracking

    Yang Zheng, Adam W Harley, Bokui Shen, Gordon Wet- zstein, and Leonidas J Guibas. Pointodyssey: A large-scale synthetic dataset for long-term point tracking. In Int. Conf. Comput. Vis. (ICCV), pages 19855–19865, 2023. 3

    2023

  55. [63]

    Event-based feature tracking with probabilistic data associa- tion

    Alex Zihao Zhu, Nikolay Atanasov, and Kostas Daniilidis. Event-based feature tracking with probabilistic data associa- tion. In IEEE Int. Conf. Robot. Autom. (ICRA), pages 4465– 4470, 2017. 3, 7, 11

    2017

  56. [64]

    From Chaos Comes Order: Ordering Event Representations for Object Recognition and Detection

    Nikola Zubic, Daniel Gehrig, Mathias Gehrig, and Davide Scaramuzza. From Chaos Comes Order: Ordering Event Representations for Object Recognition and Detection . In Int. Conf. Comput. Vis. (ICCV) , pages 12800–12810, 2023. 4 17

    2023

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.