Pith. sign in

REVIEW 3 major objections 5 minor 76 references

Multi-view Gaze Target Estimation

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Two calibrated views improve gaze target estimation, and the same pipeline can localize targets in a view that never sees the person.

desk verdict First multi-view gaze-target paper with a new dataset and a solid main result; the cross-view claim leans on an unvalidated depth-alignment step that needs a sensitivity analysis. read the letter →

arxiv 2508.05857 v1 pith:VZUM4FVK submitted 2025-08-07 cs.CV

Qiaomu Miao , Vivek Raju Golani , Jingyi Xu , Progga Paromita Dutta , Minh Hoai , Dimitris Samaras This is my paper
classification cs.CV
keywords multi-viewgazetargetestimationfollowingheadinformationaggregationuncertainty-basedselectionepipolarsceneattentioncross-viewMVGTdatasetcalibratedmulti-camerasetup
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 aims to establish that gaze target estimation — predicting where in a scene a person is looking — should not be locked to a single camera. Using a pair of synchronized, calibrated views, with one view treated as primary and the other as reference, the proposed model consistently beats single-view methods whenever the second view supplies useful face or scene information, and can even locate the target in a view that never shows the person. The authors introduce the first multi-view GTE dataset, MVGT, collected with a laser-pointer annotation protocol that yields precise targets without image artifacts. On MVGT, with the reference view showing the head, normalized target distance drops to 0.129 versus 0.141 for the best single-view baseline; in the cross-view setting, distance is 0.188 versus 0.248 for the strongest adapted baseline. If correct, the work moves GTE from a single-camera restriction to the multi-camera setups already common in public spaces.

What carries the argument

The load-bearing object is the field-of-view (FoV) heatmap, a gaze-target prior computed from a predicted 3D gaze vector $g$ and monocular depth: each pixel's vector to the eye, $V_{(u,v)}$, is scored by $M_f(u,v)=\max(0, V_{(u,v)}\cdot g/(\|V_{(u,v)}\|\|g\|))$, with exponential decay below 0.9. Three modules improve that prior. HIA cross-attends head tokens while concatenating the relative rotation $R_{21}=R_1R_2^{-1}$ into the keys/values. UGS trains the gaze estimator with the aleatoric-uncertainty loss $\mathcal{L}_{\mathrm{gaze}}=\frac{1}{2\sigma^2}(1-\hat{g}\cdot g/(\|\hat{g}\|\|g\|))+\frac{1}{2}\log\sigma^2$, then replaces the high-$\sigma$ view's gaze vector with the low-$\sigma$ vie

What would settle it

Run the cross-view pipeline in a scene where the 3D reconstruction comes from an earlier state (e.g., furniture moved) while keeping the same cameras, and measure normalized distance on gaze targets. If Dist. stays near 0.188, the depth-alignment assumption is not load-bearing; if it rises toward the 0.248 adapted-baseline level, the assumption is essential. A second check: plot angular error of predicted gaze vectors against predicted $\sigma$; UGS's selection only works while that correlation holds.

Watch

Extended reading notes

Core claim

The central claim is that two calibrated views, fused at three specific points in the pipeline, suffice to improve gaze target estimation beyond any single view. The Head Information Aggregation (HIA) module cross-attends head tokens from the other view, injecting the relative rotation between cameras so the head embedding becomes geometry-aware; the Uncertainty-based Gaze Selection (UGS) module predicts an aleatoric uncertainty $\sigma$ for each view's 3D gaze vector and replaces the less reliable prediction with the better one transformed through camera extrinsics; and the Epipolar-based Scene Attention (ESA) module lets scene tokens attend only to tokens sampled along the corresponding ep

Load-bearing premise

The load-bearing premise, which the paper itself flags in Section S14, is that cross-view GTE requires calibrated cameras and a prior 3D reconstruction, and that the RANSAC scale-and-shift fit between a per-image monocular depth map and that reconstruction yields near-absolute depth for every new frame; if the alignment is wrong, the FoV heatmap is misplaced and the cross-view prediction fails.

Editorial extensions

If this is right

  • In multi-camera environments such as lecture halls or stores, gaze following can be made more accurate than the best single-view model whenever at least one camera captures the subject's face clearly; the paper reports 0.129 versus 0.141 Dist. in that setting.
  • Gaze targets outside the field of view of the camera that sees the person become estimable: the model predicts the target in a second view using only the person's appearance, a capability single-view GTE lacks.
  • More cameras help without retraining: selecting the lowest-uncertainty pair gives Dist. 0.130 with two views, 0.121 with four, and 0.118 with six (Table S2).
  • The benefit scales with face visibility: error reduction over the single-view version is 23.7% for forward-facing, 23.2% for sideways, and 9.3% for backward heads.
  • The MVGT dataset and its laser-pointer annotation protocol give the community a calibrated, multi-view GTE benchmark with precise target labels, enabling the head/target visibility breakdown used in evaluation.

Reading between the lines

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

  • Beyond the paper: UGS's uncertainty-based replacement is a generic mechanism; the same calibrated-pair recipe (predict a per-view confidence, rotate the better prediction into the other view) could be dropped into other multi-view fusion tasks, such as multi-view action or attention prediction, wherever one view is occluded.
  • Beyond the paper: Because the cross-view path needs only a one-time scene reconstruction, a practical deployment could reconstruct an empty room once and then run continuously; the untested sensitivity is whether moving furniture or people changes the scale/shift alignment enough to break the FoV prior.
  • Beyond the paper: The ablation ties the gains to explicit camera parameters; a testable extension is to learn relative camera pose from a few views and see whether the multi-view gain survives, which would lift the calibration requirement the authors name as a limitation in Section S14.
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. The paper proposes the first multi-view gaze target estimation (GTE) method. It takes a pair of calibrated camera views, extracts head embeddings through a cross-attention Head Information Aggregation (HIA) module, selects the more reliable predicted 3D gaze vector via an Uncertainty-based Gaze Selection (UGS) module, and shares scene context through Epipolar-based Scene Attention (ESA). The method is evaluated on the new MVGT dataset, containing 13,686 synchronized images from 6 cameras across 4 scenes, with gaze targets annotated via a laser-pointer protocol. The main experiments use leave-one-scene-out cross-validation and compare against single-view GTE baselines on primary-view images. A cross-view extension predicts the gaze target in a view that does not contain the person, using the person image from another view and an absolute-depth pipeline based on Dust3R and Metric3D alignment. The paper reports consistent improvements when the reference view shows the person's head, and large gains in the cross-view setting.

Significance. If the results hold, this is a meaningful new capability and dataset: it is the first multi-view GTE method and dataset, with a novel cross-view setting that single-view methods cannot address. The method is modular, and the ablations show that each proposed module contributes to the overall improvement. The evaluation has several strengths: held-out leave-one-scene-out evaluation, a controlled comparison in which the same primary-view images are averaged over their reference pairs, a re-implementation of the Tafasca baseline with reproduced GazeFollow numbers, and supplementary analyses of uncertainty calibration and attention behavior. However, the two headline claims—(i) significant gains over single-view baselines and (ii) the cross-view capability—currently rest on a 4-scene evaluation without per-scene variance reporting and on an unvalidated absolute-depth alignment step. These issues are load-bearing for the central claims and need to be addressed before the results are fully convincing.

major comments (3)
  1. [Sec. 4.6 / S4, Eq. (9) / Table 3] The cross-view result is the paper's headline capability, but the depth-alignment step that makes it possible is never validated. Eq. (9) fits a RANSAC scale/shift (a,b) between Metric3D depth and the Dust3R reconstruction on static background pixels, with the person and calibration cube masked out; the same (a,b) is then applied to the person's monocular depth to obtain the metric 3D eye location. Nothing demonstrates that the person region shares the same scale/shift as the static background. If it does not, the transformed eye position in the primary view is systematically wrong, and the FoV heatmap is misplaced. This is not a side detail: Table 3 reports Dist 0.188 vs. 0.248 for the strongest adapted baseline, so the claimed cross-view margin depends directly on this step. The sensitivity analysis in S10 perturbs camera parameters, not depth; S14 acknowledges reliance on 3D reconstru
  2. [Sec. 5.2 / Table 1 / S1 / S12] The statistical support for the claim that the method 'significantly outperforms' single-view baselines is incomplete. The evaluation uses only four scenes (leave-one-scene-out), but no per-scene breakdown, standard error, confidence interval, or significance test is reported. The headline differences in Table 1 are small—e.g., 0.129 vs. 0.141 Dist. in the head-visible/target-visible cell—and could plausibly be driven by a single scene. Moreover, S1 shows per-scene tuning of the learning rate and the in/out loss weight (β), and the cross-view task uses learning rates of 1e-7 or 1e-8 depending on the scene. S12 shows that fixing the learning rate at one value degrades the reported numbers (0.134 vs. 0.129 for the general multi-view task; 0.199 vs. 0.188 for cross-view), so part of the reported advantage is training-protocol-dependent. Please report all four fold results, means ± standard
  3. [Sec. 4.3 / Eq. (2)] The ground-truth gaze vectors used to train the gaze estimator are computed from 'pseudo' monocular depth maps, and the paper argues that the additive shift can be ignored because Metric3D has 'low depth distortion and shift.' This is an important assumption: if a nonzero depth shift remains, the 3D vector from eye to target is not simply a scaled version of the true vector, because the shift is multiplied by the pixel-coordinate term in Eq. (5). The paper does not quantify the residual shift of Metric3D on the MVGT scenes or the effect of this residual on the FoV heatmap. Since this pseudo-depth assumption is shared with some single-view baselines (e.g., Tafasca), it may not change the comparison, but it should be validated or at least discussed quantitatively for the final heatmap computation.
minor comments (5)
  1. [Sec. 4.3] The sentence beginning 'when we use a depth estimation model that has low depth distortion and shift...' is unclear: 'V(u,v) will only be the same' should be rephrased. The mathematical reasoning about scale and shift cancellation needs to be stated precisely.
  2. [Sec. 4.4 / Fig. 6] The ESA epipolar sampling is underspecified: how are the 48 points along the epipolar line chosen (uniform in pixel coordinates? bilinear interpolation?), and are the sampled features taken from a single feature-map scale? This detail is necessary for reproducibility.
  3. [S11 / References] S11 cites '[75]' (a hedge-fund portfolio construction paper) as a transformer-architecture reference alongside [13] and [64]. This citation appears unrelated to transformer architectures and should be removed or replaced.
  4. [Sec. 5.2] The explanation 'This ensures the same total number of testing samples' is slightly misleading. Averaging over reference pairs reweights primary views that appear in different numbers of pairs; the weighting scheme should be stated explicitly.
  5. [S14] The statement that the first model is 'applicable without any assumptions' overstates its scope: the model still requires calibrated camera parameters and head bounding boxes in both views. Suggest softening this sentence.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: multi-view gains are validated on held-out manual annotations, and the cross-view depth alignment is an external geometric assumption rather than a fitted prediction.

full rationale

The paper's central claim (multi-view GTE outperforms single-view) is supported by leave-one-scene-out evaluation on the MVGT dataset with human laser-point annotations as ground truth. The modules (HIA, UGS, ESA) are trained with losses (Eq. 2, Eq. 8) whose targets are the annotated gaze coordinates and in/out labels; no test-time target information is used at inference. The cross-view pipeline (Sec. 4.6, S4) fits a per-image scale/shift between Metric3D monocular depth and Dust3R reconstruction via RANSAC on static scene regions, then transfers the eye location; this is a geometric calibration step using external depth models, not a parameter fitted to the gaze-target annotation, and the comparison with adapted baselines (Table 3) is against independent methods. Self-citations (e.g., Miao et al.) appear only as baselines or related work and are not load-bearing for the central derivation. The absence of a depth-error sensitivity analysis is a robustness limitation, not circularity.

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

The method's head-visible gains come from standard supervised training on a new dataset; the deep network contains millions of learned weights, but the explicit free parameters are the loss weights, per-scene learning rates, and ESA sampling count. The most consequential assumptions are about depth: Metric3D depth is used as if shift-free for FoV generation, and Dust3R plus RANSAC is assumed to yield absolute eye position for cross-view. These are domain assumptions with limited validation.

free parameters (4)
  • beta (in/out loss weight) = 0.05 for commons scene, 0.3 for store/kitchen/lab
    Choice of beta varies by held-out scene (S1), affecting the final loss balance.
  • learning rates for fine-tuning = 2.5e-6 lab/store, 2.5e-5 commons, 1e-5 kitchen; 1e-7/1e-8 for cross-view scenes
    Per-scene learning rates used in leave-one-scene-out validation (S1); the paper shows a small drop with a shared LR in S12.
  • ESA epipolar sampling count = 48
    Number of feature tokens sampled along the epipolar line (Sec 5.1); affects attention quality and computation.
  • loss weights alpha, lambda = 10.0, 0.1
    Weighting of heatmap MSE and gaze uncertainty losses (S1).
assumptions (5)
  • domain assumption Metric3D monocular depth has negligible shift relative to true depth, so the direction from eye to target can be recovered up to scale.
    Invoked in Sec 4.3 to justify ignoring the shift term in Eq (5); if false, FoV heatmaps misdirect the scene encoder.
  • domain assumption Dust3R, given calibrated camera parameters, produces depth close to absolute metric depth.
    Invoked in Sec 4.6 to localize the subject's eye in the primary view for cross-view GTE; not validated on the MVGT scenes.
  • domain assumption The scene is static except for the subject, so RANSAC on masked reconstruction areas yields correct scale and shift for new images.
    S4 uses calibration images to reconstruct the scene and then fits scale/shift per new input; moving subjects or changing objects break the fit.
  • domain assumption Subjects maintain their gaze at the laser-marked target after the laser is switched off.
    The dataset annotation protocol (Sec 3) assumes the point marked with laser on equals the gaze target in the paired laser-off image.
  • domain assumption Head bounding boxes from YOLOv5 and eye keypoints (or head center fallback) locate the eye accurately enough for FoV heatmap generation.
    Sec S1: if no eye is detected, the head bounding box center is used as the eye location.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-view Gaze Target Estimation." pith.science (2026). https://pith.science/paper/VZUM4FVK

@misc{pith2026250805857,
  author       = {Pith},
  title        = {Pith review of: Multi-view Gaze Target Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VZUM4FVK}},
  note         = {Machine review of arXiv:2508.05857}
}
read the original abstract

This paper presents a method that utilizes multiple camera views for the gaze target estimation (GTE) task. The approach integrates information from different camera views to improve accuracy and expand applicability, addressing limitations in existing single-view methods that face challenges such as face occlusion, target ambiguity, and out-of-view targets. Our method processes a pair of camera views as input, incorporating a Head Information Aggregation (HIA) module for leveraging head information from both views for more accurate gaze estimation, an Uncertainty-based Gaze Selection (UGS) for identifying the most reliable gaze output, and an Epipolar-based Scene Attention (ESA) module for cross-view background information sharing. This approach significantly outperforms single-view baselines, especially when the second camera provides a clear view of the person's face. Additionally, our method can estimate the gaze target in the first view using the image of the person in the second view only, a capability not possessed by single-view GTE methods. Furthermore, the paper introduces a multi-view dataset for developing and evaluating multi-view GTE methods. Data and code are available at https://www3.cs.stonybrook.edu/~cvl/multiview_gte.html

Figures

Figures reproduced from arXiv: 2508.05857 by the authors.

Figure 1
Figure 1. Benefits of multi-view GTE. Single-view GTE models [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Dataset samples and information. (a) Example images and annotations of the subject’s head location (green bounding box) and [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Images of a subject looking at the same gaze target, one [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Overall framework. Our method takes images from a pair of camera views as input. The head images are processed by the HIA [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Structures of HIA and UGS. HIA aggregates head infor [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 6
Figure 6. Figure 6: Structure of the multi-view scene encoder and ESA [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: The modified model structure for cross-view GTE. The [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Results regarding different head ori￾entations in the reference view. Our method shows a much larger improvement when the subject face is half/fully visible. 5.2. Comparison with Single-View Methods We compare our multi-view GTE method with SOTA single-view GTE methods…
Figure 9
Figure 9. Figure 9: Qualitative comparisons of our method with and without multi-view pro [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

76 extracted references · 73 canonical work pages

  1. [1]

    Social eye gaze in human-robot interaction: a review

    Henny Admoni and Brian Scassellati. Social eye gaze in human-robot interaction: a review. Journal of Human-Robot Interaction, 6(1):25–63, 2017. 1

    work page 2017

  2. [2]

    Multimae: Multi-modal multi-task masked autoen- coders

    Roman Bachmann, David Mizrahi, Andrei Atanov, and Amir Zamir. Multimae: Multi-modal multi-task masked autoen- coders. In Proceedings of the European Conference on Com- puter Vision, pages 348–367. Springer, 2022. 6

    work page 2022

  3. [3]

    Escnet: Gaze target detec- tion with the understanding of 3d scenes

    Jun Bao, Buyu Liu, and Jun Yu. Escnet: Gaze target detec- tion with the understanding of 3d scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 14126–14135, 2022. 2

    work page 2022

  4. [4]

    Unsupervised gaze representation learning from multi-view face images

    Yiwei Bao and Feng Lu. Unsupervised gaze representation learning from multi-view face images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1419–1428, 2024. 2

    work page 2024

  5. [5]

    3d pictorial structures for multiple human pose estimation

    Vasileios Belagiannis, Sikandar Amin, Mykhaylo Andriluka, Bernt Schiele, Nassir Navab, and Slobodan Ilic. 3d pictorial structures for multiple human pose estimation. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1669–1676, 2014. 7

    work page 2014

  6. [6]

    How far are we from solving the 2d & 3d face alignment problem? (and a dataset of 230,000 3d facial landmarks)

    Adrian Bulat and Georgios Tzimiropoulos. How far are we from solving the 2d & 3d face alignment problem? (and a dataset of 230,000 3d facial landmarks). In Proceedings of the IEEE/CVF International Conference on Computer Vi- sion, 2017. 7, 1

    work page 2017

  7. [7]

    Mvsnerf: Fast general- izable radiance field reconstruction from multi-view stereo

    Anpei Chen, Zexiang Xu, Fuqiang Zhao, Xiaoshuai Zhang, Fanbo Xiang, Jingyi Yu, and Hao Su. Mvsnerf: Fast general- izable radiance field reconstruction from multi-view stereo. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 14124–14133, 2021. 2

    work page 2021

  8. [8]

    Dvgaze: Dual-view gaze estima- tion

    Yihua Cheng and Feng Lu. Dvgaze: Dual-view gaze estima- tion. In Proceedings of the IEEE/CVF International Confer- ence on Computer Vision, pages 20632–20641, 2023. 2

    work page 2023

Show all 76 references
  1. [9]

    Connecting gaze, scene, and attention: Generalized attention estimation via joint modeling of gaze and scene saliency

    Eunji Chong, Nataniel Ruiz, Yongxin Wang, Yun Zhang, Agata Rozga, and James M Rehg. Connecting gaze, scene, and attention: Generalized attention estimation via joint modeling of gaze and scene saliency. In Proceedings of the European Conference on Computer Vision, pages 383–398,

  2. [10]

    Detecting attended visual targets in video

    Eunji Chong, Yongxin Wang, Nataniel Ruiz, and James M Rehg. Detecting attended visual targets in video. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5396–5406, 2020. 2, 3, 7, 6

    2020

  3. [11]

    Gaze estimation for as- sisted living environments

    Philipe Ambrozio Dias, Damiano Malafronte, Henry Medeiros, and Francesca Odone. Gaze estimation for as- sisted living environments. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages 290–299, 2020. 4, 5, 7

    2020

  4. [12]

    Fast and robust multi-person 3d pose estima- tion from multiple views

    Junting Dong, Wen Jiang, Qixing Huang, Hujun Bao, and Xiaowei Zhou. Fast and robust multi-person 3d pose estima- tion from multiple views. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 7792–7801, 2019. 2

    2019

  5. [13]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition ...

    2021

  6. [14]

    The eyes have it: the neuroethology, func- tion and evolution of social gaze

    Nathan J Emery. The eyes have it: the neuroethology, func- tion and evolution of social gaze. Neuroscience & biobehav- ioral reviews, 24(6):581–604, 2000. 1

    2000

  7. [15]

    Dual attention guided gaze target detection in the wild

    Yi Fang, Jiapeng Tang, Wang Shen, Wei Shen, Xiao Gu, Li Song, and Guangtao Zhai. Dual attention guided gaze target detection in the wild. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 11390–11399, 2021. 2

    2021

  8. [16]

    Social interactions: A first-person perspective

    Alircza Fathi, Jessica K Hodgins, and James M Rehg. Social interactions: A first-person perspective. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1226–1233. IEEE, 2012. 1

    2012

  9. [17]

    Learning to recognize daily actions using gaze

    Alireza Fathi, Yin Li, and James M Rehg. Learning to recognize daily actions using gaze. In Proceedings of the European Conference on Computer Vision, pages 314–327. Springer, 2012. 1

    2012

  10. [18]

    A meta-analysis of gaze differences to so- cial and nonsocial information between individuals with and without autism

    Thomas W Frazier, Mark Strauss, Eric W Klingemier, Emily E Zetzer, Antonio Y Hardan, Charis Eng, and Eric A Youngstrom. A meta-analysis of gaze differences to so- cial and nonsocial information between individuals with and without autism. Journal of the American Academy of Chi...

    2017

  11. [19]

    Accurate, dense, and robust multiview stereopsis

    Yasutaka Furukawa and Jean Ponce. Accurate, dense, and robust multiview stereopsis. pami, 32(8):1362–1376, 2009. 2

    2009

  12. [20]

    A modular multimodal architecture for gaze target prediction: Application to privacy-sensitive settings

    Anshul Gupta, Samy Tafasca, and Jean-Marc Odobez. A modular multimodal architecture for gaze target prediction: Application to privacy-sensitive settings. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pages 5041–5050, 2022. 2

    2022

  13. [21]

    A unified model for gaze following and social gaze prediction

    Anshul Gupta, Samy Tafasca, Naravich Chutisilp, and Jean- Marc Odobez. A unified model for gaze following and social gaze prediction. In Proceedings of the International Confer- ence on Automatic Face and Gesture Recognition, pages 1–9. IEEE, 2024. 2

    2024

  14. [22]

    MTGS: A novel framework for multi-person temporal gaze following and social gaze prediction

    Anshul Gupta, Samy Tafasca, Arya Farkhondeh, Pierre Vuil- lecard, and Jean marc Odobez. MTGS: A novel framework for multi-person temporal gaze following and social gaze prediction. In Advances in Neural Information Processing Systems, 2024. 2

    2024

  15. [23]

    Multiple view ge- ometry in computer vision

    Richard Hartley and Andrew Zisserman. Multiple view ge- ometry in computer vision . Cambridge university press,

  16. [24]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 770–778, 2016. 1, 6

    2016

  17. [25]

    Epipolar transformers

    Yihui He, Rui Yan, Katerina Fragkiadaki, and Shoou-I Yu. Epipolar transformers. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 7779–7788, 2020. 2, 5

    2020

  18. [26]

    Rotation-constrained cross-view feature fusion for multi-view appearance-based gaze estimation

    Yoichiro Hisadome, Tianyi Wu, Jiawei Qin, and Yusuke Sugano. Rotation-constrained cross-view feature fusion for multi-view appearance-based gaze estimation. In Proceed- ings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 5985–5994, 2024. 2

    2024

  19. [27]

    Denoising diffu- sion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models. Advances in Neural Information Processing Systems, 33:6840–6851, 2020. 1

    2020

  20. [28]

    We know where they are looking at from the rgb-d camera: Gaze following in 3d.IEEE Transac- tions on Instrumentation and Measurement , 71:1–14, 2022

    Zhengxi Hu, Dingye Yang, Shilei Cheng, Lei Zhou, Shichao Wu, and Jingtai Liu. We know where they are looking at from the rgb-d camera: Gaze following in 3d.IEEE Transac- tions on Instrumentation and Measurement , 71:1–14, 2022. 2

    2022

  21. [29]

    Gfie: A dataset and baseline for gaze-following from 2d to 3d in indoor environments

    Zhengxi Hu, Yuxue Yang, Xiaolin Zhai, Dingye Yang, Bo- han Zhou, and Jingtai Liu. Gfie: A dataset and baseline for gaze-following from 2d to 3d in indoor environments. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition , pages 8907–8916, 2023. 2, 3

    2023

  22. [30]

    Smart construction

    Xianjun Huang. Smart construction. https://github. com/PeterH0323/Smart_Construction. 3

  23. [31]

    Epidiff: Enhancing multi-view synthesis via localized epipolar-constrained diffusion

    Zehuan Huang, Hao Wen, Junting Dong, Yaohui Wang, Yangguang Li, Xinyuan Chen, Yan-Pei Cao, Ding Liang, Yu Qiao, Bo Dai, et al. Epidiff: Enhancing multi-view synthesis via localized epipolar-constrained diffusion. In Proceedings of the IEEE/CVF Conference on Computer Vision and...

    2024

  24. [32]

    Learnable triangulation of human pose

    Karim Iskakov, Egor Burkov, Victor Lempitsky, and Yury Malkov. Learnable triangulation of human pose. InProceed- ings of the IEEE/CVF International Conference on Com- puter Vision, pages 7718–7727, 2019. 2

    2019

  25. [33]

    Eye tracking in human- computer interaction and usability research: Ready to deliver the promises

    Robert JK Jacob and Keith S Karn. Eye tracking in human- computer interaction and usability research: Ready to deliver the promises. In The mind’s eye, pages 573–605. Elsevier,

  26. [34]

    Learning a multi-view stereo machine

    Abhishek Kar, Christian H ¨ane, and Jitendra Malik. Learning a multi-view stereo machine. Advances in Neural Informa- tion Processing Systems, 30, 2017. 2

    2017

  27. [35]

    Gaze360: Physically uncon- strained gaze estimation in the wild

    Petr Kellnhofer, Adria Recasens, Simon Stent, Wojciech Ma- tusik, and Antonio Torralba. Gaze360: Physically uncon- strained gaze estimation in the wild. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 6912–6921, 2019. 6

    2019

  28. [36]

    What uncertainties do we need in bayesian deep learning for computer vision? Advances in Neural Information Processing Systems, 30, 2017

    Alex Kendall and Yarin Gal. What uncertainties do we need in bayesian deep learning for computer vision? Advances in Neural Information Processing Systems, 30, 2017. 4, 5

    2017

  29. [37]

    Imagenet classification with deep convolutional neural net- works

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural net- works. Advances in Neural Information Processing Systems, 25, 2012. 1

    2012

  30. [38]

    Looking here or there? gaze following in 360-degree images

    Yunhao Li, Wei Shen, Zhongpai Gao, Yucheng Zhu, Guang- tao Zhai, and Guodong Guo. Looking here or there? gaze following in 360-degree images. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 3742–3751, 2021. 2

    2021

  31. [39]

    Believe it or not, we know what you are looking at! In Proceedings of the Asian Conference on Computer Vision , pages 35–50

    Dongze Lian, Zehao Yu, and Shenghua Gao. Believe it or not, we know what you are looking at! In Proceedings of the Asian Conference on Computer Vision , pages 35–50. Springer, 2018. 2

    2018

  32. [40]

    Deepgaze iie: Calibrated prediction in and out-of- domain for state-of-the-art saliency modeling

    Akis Linardos, Matthias K ¨ummerer, Ori Press, and Matthias Bethge. Deepgaze iie: Calibrated prediction in and out-of- domain for state-of-the-art saliency modeling. In Proceed- ings of the IEEE/CVF International Conference on Com- puter Vision, pages 12919–12928, 2021. 8, 6

    2021

  33. [41]

    3d gaze estimation for head-mounted devices based on visual saliency

    Meng Liu, You Fu Li, and Hai Liu. 3d gaze estimation for head-mounted devices based on visual saliency. In Proceed- ings of the IEEE/RSJ Conference on Intelligent Robots and Systems, pages 10611–10616. IEEE, 2020. 1

    2020

  34. [42]

    Zero-1-to-3: Zero-shot one image to 3d object

    Ruoshi Liu, Rundi Wu, Basile Van Hoorick, Pavel Tok- makov, Sergey Zakharov, and Carl V ondrick. Zero-1-to-3: Zero-shot one image to 3d object. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 9298–9309, 2023. 4

    2023

  35. [43]

    Patch- level gaze distribution prediction for gaze following

    Qiaomu Miao, Minh Hoai, and Dimitris Samaras. Patch- level gaze distribution prediction for gaze following. In Pro- ceedings of the IEEE/CVF Winter Conference on Applica- tions of Computer Vision, pages 880–889, 2023. 2, 7, 6

    2023

  36. [44]

    Diffusion- refined vqa annotations for semi-supervised gaze following

    Qiaomu Miao, Alexandros Graikos, Jingwei Zhang, Sounak Mondal, Minh Hoai, and Dimitris Samaras. Diffusion- refined vqa annotations for semi-supervised gaze following. In Proceedings of the European Conference on Computer Vi- sion, pages 439–457. Springer, 2024. 2

    2024

  37. [45]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis. Communications of the ACM , 65(1):99–106, 2021. 2

    2021

  38. [46]

    Dynamic 3d gaze from afar: Deep gaze estimation from temporal eye- head-body coordination

    Soma Nonaka, Shohei Nobuhara, and Ko Nishino. Dynamic 3d gaze from afar: Deep gaze estimation from temporal eye- head-body coordination. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 2192–2201, 2022. 2

    2022

  39. [47]

    Apriltag: A robust and flexible visual fiducial system

    Edwin Olson. Apriltag: A robust and flexible visual fiducial system. In Proceedings of the IEEE Conference Robotics and Automation, pages 3400–3407. IEEE, 2011. 3

    2011

  40. [48]

    3d social saliency from head-mounted cameras

    Hyun Park, Eakta Jain, and Yaser Sheikh. 3d social saliency from head-mounted cameras. Advances in Neural Informa- tion Processing Systems, 25, 2012. 1

    2012

  41. [49]

    Towards end-to-end video-based eye-tracking

    Seonwook Park, Emre Aksan, Xucong Zhang, and Otmar Hilliges. Towards end-to-end video-based eye-tracking. In Proceedings of the European Conference on Computer Vi- sion, pages 747–763. Springer, 2020. 2

    2020

  42. [50]

    Cross view fusion for 3d human pose estimation

    Haibo Qiu, Chunyu Wang, Jingdong Wang, Naiyan Wang, and Wenjun Zeng. Cross view fusion for 3d human pose estimation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4342–4351, 2019. 2 10

    2019

  43. [51]

    Where are they looking? In Advances in Neu- ral Information Processing Systems

    Adria Recasens, Aditya Khosla, Carl V ondrick, and Antonio Torralba. Where are they looking? In Advances in Neu- ral Information Processing Systems. Curran Associates, Inc.,

  44. [52]

    Following gaze in video

    Adria Recasens, Carl V ondrick, Aditya Khosla, and Anto- nio Torralba. Following gaze in video. In Proceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 1435–1443, 2017. 2, 8, 6

    2017

  45. [53]

    Nataniel Ruiz, Eunji Chong, and James M. Rehg. Fine- grained head pose estimation without keypoints. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2018. 7

    2018

  46. [54]

    Flexible cooperation be- tween human and robot by interpreting human intention from gaze information

    Kenji Sakita, Koichi Ogawara, Shinji Murakami, Kentaro Kawamura, and Katsushi Ikeuchi. Flexible cooperation be- tween human and robot by interpreting human intention from gaze information. In Proceedings of the IEEE/RSJ Con- ference on Intelligent Robots and Systems , pages 84...

    2004

  47. [55]

    Vitgaze: Gaze follow- ing with interaction features in vision transformers

    Yuehao Song, Xinggang Wang, Jingfeng Yao, Wenyu Liu, Jinglin Zhang, and Xiangmin Xu. Vitgaze: Gaze follow- ing with interaction features in vision transformers. arXiv preprint arXiv:2403.12778, 2024. 2

    2024 arXiv

  48. [56]

    Child- play: A new benchmark for understanding children’s gaze behaviour

    Samy Tafasca, Anshul Gupta, and Jean-Marc Odobez. Child- play: A new benchmark for understanding children’s gaze behaviour. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 20935–20946, 2023. 2, 5, 7, 6

    2023

  49. [57]

    Sharingan: A transformer architecture for multi-person gaze following

    Samy Tafasca, Anshul Gupta, and Jean-Marc Odobez. Sharingan: A transformer architecture for multi-person gaze following. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 2008– 2017, 2024. 2, 6

    2008

  50. [58]

    Multi- modal across domains gaze target detection

    Francesco Tonini, Cigdem Beyan, and Elisa Ricci. Multi- modal across domains gaze target detection. In Proceedings of the 2022 International Conference on Multimodal Inter- action, pages 420–431, 2022. 2

    2022

  51. [59]

    Object-aware gaze target detection

    Francesco Tonini, Nicola Dall’Asen, Cigdem Beyan, and Elisa Ricci. Object-aware gaze target detection. In Proceed- ings of the IEEE/CVF International Conference on Com- puter Vision, pages 21860–21869, 2023. 2

    2023

  52. [60]

    Al-gtd: Deep active learning for gaze target detection

    Francesco Tonini, Nicola Dall’Asen, Lorenzo Vaquero, Cig- dem Beyan, and Elisa Ricci. Al-gtd: Deep active learning for gaze target detection. In Proceedings of the 32nd ACM International Conference on Multimedia, pages 2360–2369,

  53. [61]

    End-to-end human-gaze- target detection with transformers

    Danyang Tu, Xiongkuo Min, Huiyu Duan, Guodong Guo, Guangtao Zhai, and Wei Shen. End-to-end human-gaze- target detection with transformers. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2192–2200. IEEE, 2022. 2

    2022

  54. [62]

    V oxelpose: Towards multi-camera 3d human pose estimation in wild en- vironment

    Hanyue Tu, Chunyu Wang, and Wenjun Zeng. V oxelpose: Towards multi-camera 3d human pose estimation in wild en- vironment. In Proceedings of the European Conference on Computer Vision, pages 197–212. Springer, 2020. 2

    2020

  55. [63]

    YOLOv5: A state-of-the-art real-time object de- tection system

    Ultralytics. YOLOv5: A state-of-the-art real-time object de- tection system. https://docs.ultralytics.com ,

  56. [64]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in Neural Information Processing Systems, 30, 2017. 1, 6

    2017

  57. [65]

    Gaze behavior in autism

    Fred R V olkmar and Linda C Mayes. Gaze behavior in autism. Development and Psychopathology , 2(1):61–69,

  58. [66]

    Dust3r: Geometric 3d vi- sion made easy

    Shuzhe Wang, Vincent Leroy, Yohann Cabon, Boris Chidlovskii, and Jerome Revaud. Dust3r: Geometric 3d vi- sion made easy. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20697– 20709, 2024. 2, 6

    2024

  59. [67]

    Gaze target detection based on head-local-global coordination

    Yaokun Yang and Feng Lu. Gaze target detection based on head-local-global coordination. In Proceedings of the European Conference on Computer Vision, pages 305–322. Springer, 2024. 2

    2024

  60. [68]

    Gaze target de- tection by merging human attention and activity cues

    Yaokun Yang, Yihan Yin, and Feng Lu. Gaze target de- tection by merging human attention and activity cues. In Proceedings of AAAI Conference on Artificial Intelligence , pages 6585–6593, 2024. 2

    2024

  61. [69]

    Mvs2d: Efficient multi-view stereo via attention-driven 2d convolutions

    Zhenpei Yang, Zhile Ren, Qi Shan, and Qixing Huang. Mvs2d: Efficient multi-view stereo via attention-driven 2d convolutions. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 8574– 8584, 2022. 5

    2022

  62. [70]

    Mvsnet: Depth inference for unstructured multi-view stereo

    Yao Yao, Zixin Luo, Shiwei Li, Tian Fang, and Long Quan. Mvsnet: Depth inference for unstructured multi-view stereo. In Proceedings of the European Conference on Computer Vi- sion, pages 767–783, 2018. 2

    2018

  63. [71]

    Learning to recover 3d scene shape from a single image

    Wei Yin, Jianming Zhang, Oliver Wang, Simon Niklaus, Long Mai, Simon Chen, and Chunhua Shen. Learning to recover 3d scene shape from a single image. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 204–213, 2021. 5

    2021

  64. [72]

    Metric3d: Towards zero-shot metric 3d prediction from a single image

    Wei Yin, Chi Zhang, Hao Chen, Zhipeng Cai, Gang Yu, Kaixuan Wang, Xiaozhi Chen, and Chunhua Shen. Metric3d: Towards zero-shot metric 3d prediction from a single image. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 9043–9053, 2023. 5, 6

    2023

  65. [73]

    What i see is what you see: Joint attention learning for first and third person video co-analysis

    Huangyue Yu, Minjie Cai, Yunfei Liu, and Feng Lu. What i see is what you see: Joint attention learning for first and third person video co-analysis. In Proceedings of the 27th ACM International Conference on Multimedia, pages 1358–1366,

  66. [74]

    Eth-xgaze: A large scale dataset for gaze estimation under extreme head pose and gaze variation

    Xucong Zhang, Seonwook Park, Thabo Beeler, Derek Bradley, Siyu Tang, and Otmar Hilliges. Eth-xgaze: A large scale dataset for gaze estimation under extreme head pose and gaze variation. In Proceedings of the European Confer- ence on Computer Vision , pages 365–381. Springer, 2020. 2

    2020

  67. [75]

    Hedge fund portfolio construction using polymodel theory and itransformer

    Siqiao Zhao, Zhikang Dong, Zeyu Cao, and Raphael Douady. Hedge fund portfolio construction using polymodel theory and itransformer. arXiv preprint arXiv:2408.03320 ,

    arXiv

  68. [2024]

    We extend our model to use more than 2 views (S3)

    6 11 Multi-view Gaze Target Estimation Supplementary Material Abstract In this supplementary material, we provide additional in- formation for the implementation details (S1) and statistics of the MVGT dataset (S2). We extend our model to use more than 2 views (S3). We provide...

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.