Pith. sign in

REVIEW 4 major objections 5 minor 39 references

Event-based Egocentric Human Pose Estimation in Dynamic Environment

T0 review · 4 major / 5 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 A front-facing, head-mounted event camera can estimate the wearer's 3D body mesh in dynamic scenes if moving objects are masked from the event stream first.

desk verdict New task and dataset for event-based egocentric pose, but the headline 4-of-5 win is not statistically supported and the evaluation is synthetic-only, so the practical claims are unproven. read the letter →

arxiv 2505.22007 v1 pith:S2AESAXI submitted 2025-05-28 cs.CV

Wataru Ikeda , Masashi Hatano , Ryosei Hara , Mariko Isogawa This is my paper · ORCID
classification cs.CV
keywords event-basedcameraegocentrichumanposeestimation3Dbodymeshmotionsegmentationdiffusionmodelhead-mounteddynamicenvironmentsyntheticeventdataset
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 sets out to show that a head-mounted, front-facing event camera can recover the 3D body mesh of the person wearing it, even when other people move independently through the view. The proposed framework, D-EventEgo, separates background events from dynamic-object events with a motion-segmentation module, estimates head pose from the background, and then generates the body pose with a diffusion model conditioned on that head pose. On a synthetic event dataset created from EgoBody RGB videos, the method reports gains over a re-trained RGB pipeline in four of five metrics: head orientation, head translation, acceleration, and foot skating, while joint-position error stays about the same or slightly worse. A correct result would give low-light and fast-motion applications a pose-estimation input that does not rely on visible body parts.

What carries the argument

The central mechanism is the Motion Segmentation Module, a U-Net that takes an event voxel grid (quantized into time segments and polarity bins) and outputs a binary mask of dynamic objects, trained with binary cross-entropy against masks produced by projecting ground-truth SMPL-X meshes onto the first-person views. Masking these foreground events leaves background events that the head pose module can trust, which is what makes head pose estimation accurate when other people move through the camera view. The estimated head pose then conditions the diffusion-based full-body pose generator, so the entire downstream body estimate rests on the quality of this segmentation.

What would settle it

Run the trained D-EventEgo on event data from a real front-facing, head-mounted event camera in a dim scene where one other person walks across the view, and compare head orientation, head translation, and MPJPE to the synthetic-test numbers; if accuracy degrades to baseline level or the head pose drifts, the simulator-fidelity premise that the evaluation depends on is not supported.

Watch

Extended reading notes

Core claim

D-EventEgo is proposed as the first framework for estimating a camera wearer's full 3D body mesh from a front-facing monocular event camera mounted on the head. The pipeline converts an event stream into a voxel grid, uses a U-Net motion-segmentation module to mask independently moving people, recovers head pose from the remaining background events with SLAM, gravity, and optical-flow-based networks, and feeds that head pose into a diffusion-based body pose generator. In dynamic scenes, this outperforms an EgoEgo-based baseline that was re-trained on the same event voxel grids, improving head orientation error, head translation error, acceleration, and foot skating, with MPJPE slightly worse because the baseline tends to collapse to an average pose. The paper also contributes a synthetic dataset generated from EgoBody using DVS-Voltmeter event simulation, with dynamic masks projected from ground-truth meshes.

Load-bearing premise

The entire claim rests on the assumption that synthetic event streams produced by DVS-Voltmeter from EgoBody RGB videos behave like real event camera output in the low-light and fast-motion conditions used to motivate the task; no experiment uses real event hardware or actual low-light scenes.

Editorial extensions

If this is right

  • A front-facing event camera becomes a viable low-light substitute for downward-facing body-view cameras in egocentric pose estimation, with less wearable bulk.
  • Masking independently moving objects before head-pose estimation should transfer to other event-based ego-motion and head-tracking tasks, not just body pose.
  • The diffusion body generator trained on RGB head-pose conditioning can be retrained on event-derived head poses, so existing RGB pose priors can seed event-camera systems.
  • The dataset recipe of simulating events from RGB video and projecting meshes for masks can be reused to create event training sets for new egocentric tasks.

Reading between the lines

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

  • The paper leaves implicit that the motion-segmentation module could be trained without ground-truth masks: background events obey the camera's own ego-motion, so consistency with the SLAM-estimated trajectory is a natural self-supervision signal.
  • Because MPJPE did not improve, the results suggest the body-pose diffusion prior, not head-pose accuracy, is the current bottleneck for joint-level precision; conditioning the generator on additional cues such as hand visibility would be a direct test.
  • The same segmentation-before-head-pose design may also improve RGB-based egocentric pose in dynamic scenes, since the contamination it removes is not specific to event data.
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 / 5 minor

Summary. The paper introduces D-EventEgo, the first framework for egocentric 3D human pose estimation from a front-facing, head-mounted event camera. Events are converted to voxel grids; a U-Net Motion Segmentation Module removes dynamic objects; head pose is estimated from the background voxels using DROID-SLAM, GravityNet, and HeadNet; and a conditional diffusion model (EgoEgo-style) generates full-body SMPL-X poses conditioned on the head pose. The authors construct a synthetic event dataset from EgoBody by running the DVS-Voltmeter simulator on first-person RGB videos and provide dynamic masks by projecting mocap meshes. Experiments on this dataset report that D-EventEgo outperforms an event-input-adapted EgoEgo baseline on four of five metrics (head orientation, head translation, acceleration, foot skating) while being worse on MPJPE.

Significance. If the quantitative claims held with proper statistical support and the synthetic results transferred to real event sensors, this would be a useful step toward egocentric pose estimation in low-light and fast-motion conditions, extending prior front-facing egocentric work (EgoPose, KinPoly, EgoEgo) to event data and addressing a gap left by downward-facing EventEgo3D. The paper's strengths are that it clearly defines a new task, provides a synthetic dataset with ground-truth dynamic masks, and is transparent that the comparison to EgoEgo* is effectively an ablation of the Motion Segmentation Module (Sec. 4.2). However, the main quantitative claims are currently supported only by a single run on synthetic data with small differences and no uncertainty quantification, and the motivating robustness to low light and motion blur is not empirically tested. The contribution is therefore moderate and contingent on additional evidence.

major comments (4)
  1. [Sec. 4.2, Tables 1 and 2] The central claim that D-EventEgo 'outperforms our baseline in four out of five evaluation metrics' is not statistically established. Tables 1 and 2 report single-run results with no standard deviations, confidence intervals, or significance tests. The reported differences are small (e.g., Ohead 0.293→0.282, Thead 126.6→121.8 mm, Accel 2.87→2.69 mm/s², FS 0.79→0.64 mm in Table 1; Ohead 0.286→0.277, Thead 122.57→119.23 mm in Table 2), and with 301 test sequences these gaps could easily arise from run-to-run or sequence-sampling variance. Given that the 4-of-5 claim is the paper's headline result, this missing uncertainty quantification is load-bearing. Please provide multiple training seeds with error bars, paired statistical tests, or per-sequence distributions.
  2. [Sec. 1 and Sec. 4.1] The paper motivates event cameras by their robustness in low-light conditions and high-speed motion, but every experiment uses synthetic event data generated from EgoBody RGB videos with DVS-Voltmeter. There is no validation on real event camera hardware, no low-light or fast-motion scenes beyond what the simulator produces, and no discussion of how simulator noise and dynamic range compare to real sensors. As a result, the claimed advantages for the motivating scenarios are not demonstrated. To support the central motivation, either add experiments on real event data (e.g., a small self-collected set or an existing event egocentric dataset), or substantially temper the claims and explicitly state that robustness to low light and motion blur is a hypothesis supported only by simulator assumptions.
  3. [Sec. 4.1, 4.2] The baseline EgoEgo* is EgoEgo with a modified input layer for event voxel grids, and the body pose generation module is the same diffusion backbone as EgoEgo. As the paper itself acknowledges in Sec. 4.2, the comparison 'essentially serves as an ablation study on the presence or absence of the Motion Segmentation Module.' This means the headline 'outperforms baseline' is better described as 'ablation of the Motion Segmentation Module improves four of five metrics.' Please reframe the contribution accordingly and avoid implying a head-to-head method comparison, since a true head-to-head would require baselines not sharing the same generation backbone.
  4. [Sec. 4.2] The explanation for why EgoEgo achieves lower MPJPE — 'because they often fail in estimation and tend to output an average pose' — is speculative and unsupported. If the qualitative results show D-EventEgo closer to ground truth in hand and head positions, the quantitative MPJPE regression should be explained with evidence, such as per-sequence error distributions, failure-case analysis, or a comparison of pose diversity. As written, the claim that a lower MPJPE indicates degenerate behavior needs substantiation, and the reader cannot verify it from the provided tables and figures.
minor comments (5)
  1. [Abstract] The phrase 'the first framework for this task' is ambiguous because EventEgo3D is also an event-based egocentric pose estimation method, albeit with a downward-facing camera; please clarify that 'first' refers to a front-facing head-mounted event camera.
  2. [Sec. 3.4] The definition of the full-body pose sequence is incomplete: the text states X = {x1, x2, . . . ,xT } and describes xt, but no actual equation or loss function for the diffusion model is provided. Please add the complete formulation or an explicit reference to the EgoEgo equations.
  3. [Sec. 3.2, Eq. (1)] The BCE loss in Eq. (1) sums over H and W only, but the mask is described as binary at each voxel position in a T×H×W voxel grid. Please clarify whether the mask is defined per frame (summed/averaged over T) or whether the loss should also sum over the temporal dimension.
  4. [Sec. 4.1] The relationship between sequence length (150 frames), the voxelization frame rate (30 FPS), the number of temporal segments T, and the number of time bins B is not spelled out. Please define T and B explicitly in relation to the 150-frame sequences.
  5. [Fig. 4 and Fig. 5] Some symbols in Fig. 4 (e.g., 'Voxel Grid !', 'Head Poses "', 'Full Body Poses #') appear as raw notation and are difficult to read; please format these consistently. In Fig. 5, the label 'GTw/ Motion Segmentation' is awkward and should be reworded.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the training/evaluation splits are separate and the baseline comparison is an acknowledged ablation, not a self-referential prediction.

full rationale

The paper's derivation chain is self-contained. The Motion Segmentation Module is trained with ground-truth dynamic masks obtained by projecting EgoBody SMPL-X meshes into the first-person view, and then evaluated on held-out test sequences; the head-pose and body-pose modules are likewise trained and tested on separate splits. The baseline 'EgoEgo*' is the same external EgoEgo backbone adapted to event input by a modified input layer, and the paper explicitly states that the comparison 'essentially serves as an ablation study on the presence or absence of the Motion Segmentation Module' (Sec. 4.2). There is no equation that defines the predicted quantity in terms of the fitting target, no fitted parameter is renamed as a prediction, and no load-bearing argument depends on a self-citation: EgoEgo, DROID-SLAM, DVS-Voltmeter, and the other cited components are prior external work. The absence of error bars or repeated runs is a statistical-reporting limitation rather than a circularity, because the reported improvements are not forced by construction. Accordingly, no circular step is present.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the assumption that synthetic events from DVS-Voltmeter faithfully represent real event camera behavior, and that EgoBody's ground truth poses are accurate. The method also inherits EgoEgo's assumption that head pose is a sufficient condition for body pose, and uses DROID-SLAM outside its intended input domain. No new physical entities are proposed; the free parameters are hand-chosen sequence length, bin count, and frame rate that shape the input representation.

free parameters (3)
  • Sequence length T = 150 frames
    Chosen to limit SLAM error accumulation; affects all metrics and is a hand-selected design choice.
  • Number of time bins B = 3
    Voxelization parameter; hand-chosen, controls temporal granularity of the input representation.
  • Frame rate = 30 FPS
    Defines temporal segmentation of event cloud; arbitrary choice, influences input structure.
assumptions (5)
  • domain assumption Synthetic events from DVS-Voltmeter approximate real event camera output, including low-light behavior
    The entire evaluation uses synthetic events from EgoBody RGB video; no real event sensor validation, and low-light scenes are not simulated.
  • domain assumption EgoBody ground truth SMPL-X poses and projected masks are accurate enough to supervise training and evaluation
    Dynamic masks are created by projecting GT meshes; errors in EgoBody GT propagate to both segmentation training and evaluation.
  • domain assumption Head pose estimated from egocentric event data is a sufficient condition signal for full-body pose generation
    Adopted from EgoEgo; no analysis of conditioning ambiguity or failure cases.
  • domain assumption Pre-trained DROID-SLAM generalizes to event voxel grids despite being trained on frame-based data
    Used as-is for camera pose estimation; only a brief claim that domain gap is reduced, no quantitative check.
  • standard math Standard deep learning assumptions (U-Net, BCE loss, diffusion training) are reliable
    Common practice in the field.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Event-based Egocentric Human Pose Estimation in Dynamic Environment." pith.science (2026). https://pith.science/paper/S2AESAXI

@misc{pith2026250522007,
  author       = {Pith},
  title        = {Pith review of: Event-based Egocentric Human Pose Estimation in Dynamic Environment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S2AESAXI}},
  note         = {Machine review of arXiv:2505.22007}
}
read the original abstract

Estimating human pose using a front-facing egocentric camera is essential for applications such as sports motion analysis, VR/AR, and AI for wearable devices. However, many existing methods rely on RGB cameras and do not account for low-light environments or motion blur. Event-based cameras have the potential to address these challenges. In this work, we introduce a novel task of human pose estimation using a front-facing event-based camera mounted on the head and propose D-EventEgo, the first framework for this task. The proposed method first estimates the head poses, and then these are used as conditions to generate body poses. However, when estimating head poses, the presence of dynamic objects mixed with background events may reduce head pose estimation accuracy. Therefore, we introduce the Motion Segmentation Module to remove dynamic objects and extract background information. Extensive experiments on our synthetic event-based dataset derived from EgoBody, demonstrate that our approach outperforms our baseline in four out of five evaluation metrics in dynamic environments.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 36 canonical work pages

  1. [1]

    Event-based Egocentric Human Pose Estimation in Dynamic Environment

    INTRODUCTION Estimating 3D human pose from egocentric vision is a crucial task in various applications, such as sports motion analysis and VR/AR applications. Human pose estimation using a front-facing monocular camera, as shown in Fig. 1, is one of the most commonly adopted camera setups for captur- ing egocentric videos [ 1, 2, 3]. Existing works that e...

    work page Pith review arXiv 2025

  2. [2]

    In recent years, sig- nificant advancements have been made in egocentric human pose estimation

    RELATED WORKS Egocentric Human Pose Estimation. In recent years, sig- nificant advancements have been made in egocentric human pose estimation. Based on the orientation of a head-mounted camera, existing approaches can be primarily categorized into two groups [ 6]. One involves methods that mount the camera in a front-facing orientation, which does not ass...

    work page

  3. [3]

    The input consists of an event cloud E, which is composed of N event points ek = (xi, yi, ti, pi) captured by a monocular event-based camera affixed frontally to the subject’s head

    METHOD We address the task of estimating egocentric 3D human poses from a front-facing event-based camera mounted on the head in dynamic environments. The input consists of an event cloud E, which is composed of N event points ek = (xi, yi, ti, pi) captured by a monocular event-based camera affixed frontally to the subject’s head. x, y rep- resent the 2D p...

    work page

  4. [4]

    Experimental Setup Datasets

    EXPERIMENTS 4.1. Experimental Setup Datasets. Since we tackle a new task (i.e., event-based ego- centric human pose estimation with front-facing camera), we constructed an original dataset. Specifically, we created a new dataset using the EgoBody [ 16] dataset. EgoBody consists of a combination of first-person RGB videos, SMPL-X pose data of the individuals...

    work page

  5. [5]

    By removing the event data of the person appearing in the reflection using the Motion Table 2: The impact of motion segmentation module on head pose estimation on test sets. Method Ohead ↓ Thead [mm] ↓ Ours w/o Motion Segmentation 0.286 122.57 Ours 0.277 119.23 EgoEgo w/ GT mask 0.275 118.77 Segmentation Module, we demonstrated an improvement in the accura...

    work page

  6. [6]

    To address the issue of dynamic objects reflected in the front-facing egocentric camera, we introduced a Motion Segmentation Module to remove dynamic objects

    CONCLUSION This paper tackled a new task of estimating the full-body pose of the camera wearer using a front-facing event-based egocentric camera. To address the issue of dynamic objects reflected in the front-facing egocentric camera, we introduced a Motion Segmentation Module to remove dynamic objects. Since this is a new task, we created an original dat...

    work page

  7. [7]

    Ego-pose estimation and forecasting as real-time pd control,

    Y e Y uan and Kris Kitani, “Ego-pose estimation and forecasting as real-time pd control,” in ICCV, 2019

    work page 2019

  8. [8]

    Dynamics-regulated kinematic policy for egocen- tric pose estimation,

    Zhengyi Luo, Ryo Hachiuma, Y e Y uan, and Kris Ki- tani, “Dynamics-regulated kinematic policy for egocen- tric pose estimation,” NeurIPS, 2021

    work page 2021

Show all 39 references
  1. [9]

    Ego-body pose estimation via ego-head pose estimation,

    Jiaman Li, Karen Liu, and Jiajun Wu, “Ego-body pose estimation via ego-head pose estimation,” in CVPR, 2023

    2023

  2. [10]

    Event-based vision: A survey,

    Guillermo Gallego, Tobi Delbr ¨uck, Garrick Orchard, Chiara Bartolozzi, Brian Taba, Andrea Censi, Stefan Leutenegger, Andrew J Davison, J ¨org Conradt, Kostas Daniilidis, et al., “Event-based vision: A survey,” IEEE TPAMI, 2019

    2019

  3. [11]

    Eventego3d: 3d human motion capture from egocentric event streams,

    Christen Millerdurai, Hiroyasu Akada, Jian Wang, Diogo Luvizon, Christian Theobalt, and Vladislav Golyanik, “Eventego3d: 3d human motion capture from egocentric event streams,” in CVPR, 2024

    2024

  4. [12]

    A Survey on 3D Egocentric Human Pose Estimation,

    Md Mushfiqur Azam and Kevin Desai, “A Survey on 3D Egocentric Human Pose Estimation,” in CVPRW, 2024

    2024

  5. [13]

    Estimating Body and Hand Motion in an Ego-sensed World,

    Brent Yi, Vickie Y e, Maya Zheng, Lea M ¨uller, Geor- gios Pavlakos, Yi Ma, Jitendra Malik, and Angjoo Kanazawa, “Estimating Body and Hand Motion in an Ego-sensed World,” arXiv preprint arXiv:2410.03665 , 2024

    2024 arXiv

  6. [14]

    Selfpose: 3d egocentric pose esti- mation from a headset mounted camera,

    Denis Tome, Thiemo Alldieck, Patrick Peluse, Gerard Pons-Moll, Lourdes Agapito, Hernan Badino, and Fer- nando De la Torre, “Selfpose: 3d egocentric pose esti- mation from a headset mounted camera,” IEEE TPAMI, 2020

    2020

  7. [15]

    Estimating egocentric 3d human pose in global space,

    Jian Wang, Lingjie Liu, Weipeng Xu, Kripasindhu Sarkar, and Christian Theobalt, “Estimating egocentric 3d human pose in global space,” in CVPR, 2021

    2021

  8. [16]

    Egoglass: Egocentric-view human pose estimation from an eyeglass frame,

    Dongxu Zhao, Zhen Wei, Jisan Mahmud, and Jan- Michael Frahm, “Egoglass: Egocentric-view human pose estimation from an eyeglass frame,” in 3DV, 2021

    2021

  9. [17]

    Unrealego: A new dataset for robust egocentric 3d hu- man motion capture,

    Hiroyasu Akada, Jian Wang, Soshi Shimada, Masaki Takahashi, Christian Theobalt, and Vladislav Golyanik, “Unrealego: A new dataset for robust egocentric 3d hu- man motion capture,” in ECCV, 2022

    2022

  10. [18]

    Scene- aware egocentric 3d human pose estimation,

    Jian Wang, Diogo Luvizon, Weipeng Xu, Lingjie Liu, Kripasindhu Sarkar, and Christian Theobalt, “Scene- aware egocentric 3d human pose estimation,” in CVPR, 2023

    2023

  11. [19]

    Ego3dpose: Capturing 3d cues from binocular egocentric views,

    Taeho Kang, Kyungjin Lee, Jinrui Zhang, and Y oungki Lee, “Ego3dpose: Capturing 3d cues from binocular egocentric views,” in SIGGRAPH Asia, 2023

    2023

  12. [20]

    3d human pose perception from egocentric stereo videos,

    Hiroyasu Akada, Jian Wang, Vladislav Golyanik, and Christian Theobalt, “3d human pose perception from egocentric stereo videos,” in CVPR, 2024

    2024

  13. [21]

    Simpleego: Predicting probabilistic body pose from egocentric cameras,

    Hanz Cuevas-V elasquez, Charlie Hewitt, Sadegh Aliak- barian, and Tadas Baltru ˇsaitis, “Simpleego: Predicting probabilistic body pose from egocentric cameras,” in 3DV, 2024

    2024

  14. [22]

    Egobody: Human body shape and motion of inter- acting people from head-mounted devices,

    Siwei Zhang, Qianli Ma, Y an Zhang, Zhiyin Qian, Taein Kwon, Marc Pollefeys, Federica Bogo, and Siyu Tang, “Egobody: Human body shape and motion of inter- acting people from head-mounted devices,” in ECCV, 2022

    2022

  15. [23]

    Aria dig- ital twin: A new benchmark dataset for egocentric 3d machine perception,

    Xiaqing Pan, Nicholas Charron, Y ongqian Y ang, Scott Peters, Thomas Whelan, Chen Kong, Omkar Parkhi, Richard Newcombe, and Y uheng Carl Ren, “Aria dig- ital twin: A new benchmark dataset for egocentric 3d machine perception,” in ICCV, 2023

    2023

  16. [26]

    Droid-slam: Deep vi- sual slam for monocular, stereo, and rgb-d cameras,

    Zachary Teed and Jia Deng, “Droid-slam: Deep vi- sual slam for monocular, stereo, and rgb-d cameras,” NeurIPS, 2021

    2021

  17. [28]

    Expressive body cap- ture: 3d hands, face, and body from a single image,

    Georgios PPavlakos, V asileios Choutas, Nima Ghor- bani, Timo Bolkart, Ahmed AA Osman, Dimitrios Tzionas, and Michael J Black, “Expressive body cap- ture: 3d hands, face, and body from a single image,” in CVPR, 2019

    2019

  18. [31]

    Nemf: Neural motion fields for kinematic animation,

    Chengan He, Jun Saito, James Zachary, Holly Rush- meier, and Yi Zhou, “Nemf: Neural motion fields for kinematic animation,” NeurIPS, 2022. EVENT-BASED EGOCENTRIC HUMAN POSE ESTIMA TION IN DYNAMIC ENVIRONMENT Supplementary Material Contents 1 Overview of the Supplementary Materi...

    2022

  19. [32]

    We provide a video demo to obtain more qualitative results

    OVERVIEW OF THE SUPPLEMENTARY MA TERIAL The supplementary material includes details on the model architecture, implementation, baselines, and synthetic dataset. We provide a video demo to obtain more qualitative results

  20. [33]

    U-Net consists of an encoder and a decoder, with the input being voxel grid [ 2] and the output being a segmentation mask

    MODEL ARCHITECTURE Motion Segmentation Module The Motion Segmentation Module is a network based on U-Net [ 1]. U-Net consists of an encoder and a decoder, with the input being voxel grid [ 2] and the output being a segmentation mask. The encoder extracts features by reducing t...

  21. [34]

    The training was conducted for 100 epochs with a batch size of 32, a learning rate of 1.0×10−5, and AdamW [ 6] as the optimization algorithm

    IMPLEMENTA TION DETAILS Motion Segmentation Module We utilized an NVIDIA GeForce RTX 4090 GPU, and the Motion Segmentation Module training process reached convergence in roughly 23 hours. The training was conducted for 100 epochs with a batch size of 32, a learning rate of 1.0...

  22. [35]

    First, event data was generated from the RGB first- person view videos of the EgoBody dataset using the event simulator DVS-V oltmeter [ 8]

    SYNTHETIC DA TASET Synthetic dataset was created based on the EgoBody [ 7] dataset. First, event data was generated from the RGB first- person view videos of the EgoBody dataset using the event simulator DVS-V oltmeter [ 8]. Subsequently, voxelization [ 2] was performed to gene...

  23. [36]

    The video demo com- pares the results of our baseline, which inputs event data into EgoEgo [ 5], and our proposed method, D-EventEgo

    VIDEO QUALITA TIVE EV ALUA TION To conduct a qualitative evaluation of the proposed method, we provide a video demo. The video demo com- pares the results of our baseline, which inputs event data into EgoEgo [ 5], and our proposed method, D-EventEgo. The scenes include three d...

  24. [37]

    U-net: Convolutional networks for biomedical image segmentation,

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox, “U-net: Convolutional networks for biomedical image segmentation,” in MICCAI, 2015

    2015

  25. [38]

    Unsupervised event-based optical flow using motion compensation,

    Alex Zihao Zhu, Liangzhe Y uan, Kenneth Chaney, and Kostas Daniilidis, “Unsupervised event-based optical flow using motion compensation,” in ECCVW, 2018

    2018

  26. [39]

    Droid-slam: Deep visual slam for monocular, stereo, and rgb-d cameras,

    Zachary Teed and Jia Deng, “Droid-slam: Deep visual slam for monocular, stereo, and rgb-d cameras,” NeurIPS, 2021

    2021

  27. [40]

    Deep residual learning for image recognition,

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun, “Deep residual learning for image recognition,” in CVPR, 2016

    2016

  28. [41]

    Ego-body pose es- timation via ego-head pose estimation,

    Jiaman Li, Karen Liu, and Jiajun Wu, “Ego-body pose es- timation via ego-head pose estimation,” in CVPR, 2023

    2023

  29. [42]

    Decoupled weight decay regularization,

    Ilya Loshchilov and Frank Hutter, “Decoupled weight decay regularization,” arXiv preprint arXiv:1711.05101 , 2017

    2017 arXiv

  30. [43]

    Egobody: Human body shape and motion of interacting people from head-mounted devices,

    Siwei Zhang, Qianli Ma, Y an Zhang, Zhiyin Qian, Taein Kwon, Marc Pollefeys, Federica Bogo, and Siyu Tang, “Egobody: Human body shape and motion of interacting people from head-mounted devices,” in ECCV, 2022

    2022

  31. [44]

    DVS-V oltmeter: Stochastic process-based event simu- lator for dynamic vision sensors,

    Songnan Lin, Y e Ma, Zhenhua Guo, and Bihan Wen, “DVS-V oltmeter: Stochastic process-based event simu- lator for dynamic vision sensors,” in ECCV, 2022

    2022

Pith tools

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