Pith. sign in

REVIEW 4 major objections 5 minor 64 references

Understanding Human Gaze Communication by Spatio-Temporal Graph Reasoning

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper introduces the first dataset and model that read human gaze communication in social video at two levels, and reports that its graph reasoning model reaches 55.02% top-1 accuracy on atomic-level prediction.

desk verdict The dataset is the contribution; the benchmark numbers are plausible but need inter-annotator agreement and statistical grounding before VACATION becomes a stable reference. read the letter →

arxiv 1909.02144 v1 pith:T47FWLD2 submitted 2019-09-04 cs.CV

Lifeng Fan , Wenguan Wang , Siyuan Huang , Xinyu Tang , Song-Chun Zhu This is my paper · ORCID
classification cs.CV
keywords gazecommunicationsocialvideounderstandinggraphneuralnetworkspatio-temporalreasoningVACATIONdatasetjointattentionatomicclassificationevent-levelrecognition
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 introduces a new task—understanding human gaze communication in third-person social video—and argues that the task splits naturally into six atomic gaze types (single, mutual, avert, refer, follow, share) and five event-level types (non-communicative, mutual gaze, gaze aversion, gaze following, joint attention). To support the task it contributes VACATION, a dataset of 300 social video clips with annotations of faces, objects, attention, and both levels of gaze labels. The paper's central experimental claim is that a spatio-temporal graph neural network, which treats people and the scene as nodes and learns directed gaze relations as edges, reaches 55.02% top-1 average accuracy on atomic-level prediction and outperforms the compared baselines, while the event network improves event-level prediction to 37.1% with predicted atomic labels and 55.9% with ground-truth atomic labels. A sympathetic reader would care because the paper supplies a structured vocabulary and a benchmark for a socially significant behavior that vision systems currently treat only indirectly through saliency or object attention.

What carries the argument

The load-bearing mechanism is a complete directed social graph built per frame. Nodes are the human faces in the frame plus one scene node; directed edges stand for all possible gaze relations; connectivity weights form an adjacency matrix $A \in [0,1]^{|V| \times |V|}$. The model alternates edge updates, structure updates of $A$, and node updates over several message-passing iterations, then applies an LSTM to each human node over time and reads out atomic labels. A separate event network encodes the transition and frequency vectors of the predicted atomic sequence and decodes the event label. The main version of the model learns $A$ explicitly under supervision of the annotated attention relations, while an implicit variant treats $A$ as an attention mechanism.

What would settle it

Re-annotate the VACATION test set independently with a new team using the same annotation instructions and measure agreement on atomic gaze labels; if agreement is low (say, below 80%), the 55.02% top-1 accuracy is not a stable property of the model but an artifact of one annotation pass. Alternatively, train the full model using random permutations of the ground-truth adjacency matrices; if top-1 accuracy stays near 55%, the explicitly learned communication structure is not the load-bearing component.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that gaze communication in social video can be modeled as inference over a spatio-temporal social graph. Each frame gives a complete directed graph whose nodes are human faces plus a scene node; directed edges represent possible gaze interactions, and an adjacency matrix encodes who is attending to whom. The model iteratively updates edge and node representations through message passing, adds an LSTM over time for each human node, and then reads out atomic gaze labels; a separate encoder-decoder network composes the predicted atomic sequence into an event-level label. On VACATION, the full model achieves 55.02% top-1 average accuracy on atomic-level prediction and 37.1% on event-level prediction using predicted atomic labels, improving substantially over the baselines the paper compares. The authors also claim to be the first to formulate this task, so the dataset and this accuracy are offered as the reference point for all later work.

Load-bearing premise

The load-bearing premise is that the ground-truth adjacency annotations—who is looking at whom, and the six category labels—are complete and accurate; if those annotations miss or mislabel common gaze behavior, the reported accuracy partly measures the annotation conventions rather than gaze understanding.

Editorial extensions

If this is right

  • If correct, any third-person social video can be parsed into a structured set of "who looks at whom" relations, and those relations are the atoms of social communication.
  • The 55.02% atomic accuracy becomes a reference number against which later models can be compared on the same VACATION split without redefining the task.
  • Event-level prediction, such as joint attention, is improved by grounding it in predicted atomic gazes, suggesting that hierarchical temporal composition is a viable route for social event understanding.
  • Systems for robotics and virtual reality can use the learned graph structure to interpret or generate gaze behavior rather than treating gaze as a low-level feature.
  • Because the adjacency matrix can also be learned implicitly, the framework offers a path toward gaze-relation discovery when explicit attention annotations are unavailable, though the paper's implicit variant performs noticeably worse.

Reading between the lines

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

  • Beyond the paper: if the six-category taxonomy proves stable under re-annotation, it could become a standard annotation scheme for social gaze, comparable to action labels in video datasets.
  • Beyond the paper: the explicit adjacency supervision is likely the main source of the reported gains; a natural test is whether a self-supervised adjacency from estimated gaze direction reaches similar accuracy, which would let the model scale to videos without manual attention labels.
  • Beyond the paper: the same graph formulation might transfer to audio-visual scenes where gaze is partly occluded, using head pose and object affordances as node features.
  • Beyond the paper: the event network's use of transition and frequency vectors suggests that event labels are largely determined by the order and duration of atomic gazes; if true, counting and transitions are the discriminative signal, and more sophisticated temporal models may add little.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 a new task, gaze communication understanding in social videos, at two levels: atomic-level (six classes: single, mutual, avert, refer, follow, share) and event-level (five classes: Non-communicative, Mutual Gaze, Gaze Aversion, Gaze Following, Joint Attention). It presents VACATION, a dataset of 300 video clips with face/object bounding boxes, human attention annotations, and atomic- and event-level labels. The proposed method is a spatio-temporal graph neural network that performs iterative message passing, learns an adjacency matrix over human and scene nodes, applies an LSTM over time, and feeds atomic predictions to an encoder-decoder event network. Experiments report 55.02% top-1 accuracy for atomic-level prediction and 55.9% for event-level prediction with ground-truth atomic labels, alongside several baselines and ablations.

Significance. If the dataset annotations are reliable, VACATION is a potentially valuable resource for a genuinely new problem, and the proposed graph-based model is a reasonable first baseline. The paper contributes a new task formulation, a nontrivial annotation effort (over 96k frames, 206k face boxes, 164k gaze communication relations), and a public code repository, all of which are strengths. However, the central benchmark claims currently rest on unverified annotation reliability and on comparisons without error bars or significance tests. The event-level experiment is partly a check on a handcrafted composition rule rather than an independent prediction task. With additional annotation-quality analysis, statistical evaluation, and a rule-based event baseline, the paper could serve as a solid benchmark reference; in its present form the quantitative claims are not yet established to journal standard.

major comments (4)
  1. [§3.2, Table 1] No inter-annotator agreement is reported for the six atomic classes, the five event classes, or the directed adjacency relations A that supervise the graph. The text only states that two volunteers annotated separately and a specialist adjudicated differences. Since the dataset is the main contribution and the reported 55.02% accuracy is the central benchmark claim, label reliability is load-bearing. The ambiguity is visible in Table 1: Gaze Following events contain 39.26% single and 49.41% share atomic labels, with 'follow' itself only 10.71%, so boundary cases such as follow vs. share vs. single are likely frequent. The authors should report kappa or equivalent agreement measures, discuss the most confusable class pairs, and quantify how adjudication changed the labels.
  2. [§5.2, Table 3] The full model is reported as 'Ours-full (iteration 2)', but the paper does not state a validation-based rule for selecting the number of message-passing iterations. The iteration ablation shows Ours-iteration1 at 33.67%, Ours-iteration2 at 55.02%, Ours-iteration3 at 44.52%, and Ours-iteration4 at 48.35%, which is a large, non-monotonic variation. Without a clear criterion such as best validation accuracy, the choice of iteration 2 appears post hoc. Additionally, no error bars or significance tests are given for any table entry, so the claim of 'significant' improvement over baselines is not statistically supported. The authors should specify the model-selection procedure, report mean and standard deviation over multiple runs, and include paired significance tests for the main comparisons.
  3. [§1, §4.2, Table 4] The event-level evaluation is partly circular. In Section 1 the event labels are explicitly defined as temporal compositions of atomic labels (e.g., Gaze Aversion starts from mutual then avert to single; Joint Attention consists of mutual, refer, follow, and mutual). The event network with ground-truth atomic labels therefore mainly tests whether the encoder-decoder learns a known handcrafted mapping. The paper should include a deterministic rule-based baseline that applies the composition definitions directly to ground-truth atomic sequences and compare against it. It is also notable that Ours-w. GT (55.9%) is only slightly above FC-w. GT (52.6%), and no significance test is provided for this difference.
  4. [§5.2, Table 3] The claim that the full model shows 'good and balanced performance for each atomic type' is not supported by the reported precision and F1 values. From the table, Ours-full has precision 22.10% and F1 26.17% for 'single', implying recall of roughly 32%, and for 'follow' precision 32.83% with F1 18.05%, implying recall of roughly 12%. These are low and highly imbalanced figures. The paper should report per-class recall, class-specific confusion analysis, and a discussion of why rare classes such as 'follow' are poorly recovered, rather than relying on average top-1 accuracy.
minor comments (5)
  1. [§4.2, Eq. (2)] The activation function σ in Eq. (2) should be explicitly identified as a sigmoid to ensure that the adjacency entries a_{v,w} lie in [0,1] as defined in §4.1.
  2. [§5.2] There is a typo: 'Restnet50' should be 'ResNet50'.
  3. [Figure 4 caption] The caption refers to 'the last paragraph in §4.1' for detailed descriptions, but the detailed architecture descriptions appear in §4.2; please correct the cross-reference.
  4. [§4.2, Node/Edge Feature Initialization] The compression of 4096-d fc7 features to 6-d after concatenating 6-d position information is described only briefly; a sentence explaining why 6-d is sufficient would improve clarity.
  5. [§3.2, Table 2] The abbreviation GCR is used in Table 2 and defined only in the table caption; please define it in the main text as well.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial circularity: the event-level '55.9% with ground-truth atomics' result is the network learning the paper's own definition of events as temporal compositions of atomics; the central atomic benchmark is independent.

  1. self definitional [§1 (Fig. 1 caption) and §4.1 'Event Network' and Table 4]
    "Event-level gaze communication refers to high-level, complex social communication events, including Non-communicative, Mutual Gaze, Gaze Aversion, Gaze Following and Joint Attention. Each gaze communication event is a temporal composition of some atomic-level gaze communications ... we pre-process the input atomic gaze sequence into two vectors: i) the transition vector that records each transition from one category of atomic gaze to another, and ii) the frequency vector that computes the frequency of each atomic type. ..."

    The paper defines the five event categories as temporal compositions of the six atomic labels (e.g., Gaze Following is follow+share without mutual; Joint Attention is mutual->refer->follow->mutual). The event network's input is exactly the atomic gaze sequence summarized by transition and frequency vectors. Consequently, the reported 55.9% event accuracy 'with ground truth atomic labels' is not an independent event-recognition result; it is a learned reconstruction of the composition rule that was used to define the event taxonomy. The result does not reduce to a tautology because the network is trained on separately annotated event labels and some event descriptions are 'typical' rather than exact, but the definitional dependence makes the GT-atomic event result partially circular.

full rationale

The primary atomic-level contribution is self-contained: the model is trained on held-out annotations (atomic labels plus supervised adjacency matrix A) and evaluated on a separate test set; no fitted parameter is renamed as a prediction, and no load-bearing uniqueness theorem or ansatz is imported from the authors' prior work. The self-citations [17] and [46] appear only in related-work/method-inspiration contexts and do not carry the argument. The only real circularity concern is the event-level 'with ground-truth atomic labels' result: because the paper defines each event as a temporal composition of atomic gazes, the event network's input (transition+frequency of the atomic sequence) contains the defining components of the target, so the 55.9% figure is partly a consistency check on the handcrafted taxonomy rather than an independent event-level prediction. The 'w/o GT' event result (37.1%) and the atomic benchmark remain independent. The absence of inter-annotator agreement statistics is a correctness/annotation-reliability concern, not circularity, and does not raise the circularity score beyond this partial definitional dependence.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central benchmark claim rests on the annotation taxonomy and the availability of ground-truth gaze relations. The model uses standard neural network training, so trained weights are not listed; the listed hyperparameters are hand-chosen and the domain assumptions are load-bearing.

free parameters (2)
  • number of message passing iterations N = 2 (selected among 1-4)
    The full model is reported as Ours-iteration 2 in Table 3; iterations 1, 3, and 4 give different accuracies, indicating sensitivity to this hand-chosen hyperparameter.
  • LSTM sequence length = 5 frames
    Set according to dataset statistics in Section 5.1; no sensitivity analysis is provided.
assumptions (4)
  • domain assumption The six atomic gaze categories and five event categories are exhaustive and mutually exclusive for the annotated social scenes.
    The entire label space, loss functions, and metrics depend on this taxonomy defined in Section 1; no inter-annotator agreement statistics are reported.
  • domain assumption Ground-truth gaze attention relations (adjacency matrix A) are accurately annotated.
    Section 4.2 supervises A explicitly with cross-entropy; annotation cross-validation is claimed in Section 3.2 but no agreement numbers are given.
  • domain assumption The event-level labels are temporal compositions of atomic labels as described in Section 1.
    This compositional definition is used to justify the event network design and to interpret the GT-atomic-label results in Table 4.
  • domain assumption ResNet50 pretrained on ImageNet provides a useful feature space for gaze communication.
    Section 4.2 initializes node features from ResNet50 fc7; no analysis of domain shift is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Understanding Human Gaze Communication by Spatio-Temporal Graph Reasoning." pith.science (2026). https://pith.science/paper/T47FWLD2

@misc{pith2026190902144,
  author       = {Pith},
  title        = {Pith review of: Understanding Human Gaze Communication by Spatio-Temporal Graph Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T47FWLD2}},
  note         = {Machine review of arXiv:1909.02144}
}
read the original abstract

This paper addresses a new problem of understanding human gaze communication in social videos from both atomic-level and event-level, which is significant for studying human social interactions. To tackle this novel and challenging problem, we contribute a large-scale video dataset, VACATION, which covers diverse daily social scenes and gaze communication behaviors with complete annotations of objects and human faces, human attention, and communication structures and labels in both atomic-level and event-level. Together with VACATION, we propose a spatio-temporal graph neural network to explicitly represent the diverse gaze interactions in the social scenes and to infer atomic-level gaze communication by message passing. We further propose an event network with encoder-decoder structure to predict the event-level gaze communication. Our experiments demonstrate that the proposed model improves various baselines significantly in predicting the atomic-level and event-level gaze

Figures

Figures reproduced from arXiv: 1909.02144 by the authors.

Figure 1
Figure 1. We study human gaze communication dynamics in two hierarchical levels: atomic-level and event-level. Atomic-level gaze communication describes the fine-grained structures in human gaze interactions, i.e., single, mutual, avert, refer, follow and share (as shown in the left part). Event-level gaze communication refers to high-level, complex social communication events, including Non-communicative, Mutual Gaze, Gaze A… view at source ↗
Figure 2
Figure 2. Example frames and annotations of our VACATION dataset, showing that our dataset covers rich gaze communication behaviors, diverse general social scenes, different cultures, etc. It also provides rich annotations, i.e., human face and object bboxes, gaze communication structures and labels. Human faces and related objects are marked by boxes with the same color of corresponding communication labels. White lines link… view at source ↗
Figure 3
Figure 3. Illustration of the proposed spatio-temporal reason￾ing model for gaze communication understanding. Given an input social video sequence (a), for each frame, a spatial reasoning pro￾cess (b) is first performed for simultaneously capturing gaze com￾munication relations (social graph structure) and updating node representations through message propagation. Then, in (c), a temporal reasoning process is applied for each… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Detailed architecture of the proposed spatio-temporal reasoning model for gaze communication understanding. See the last paragraph in §4.1 for detailed descriptions. reasoning. With the updated human node representations {yv ∈ R V }v∈V\s from our message passing based …
Figure 5
Figure 5. Figure 5: Qualitative results of atomic-level gaze communication prediction. Correctly inferred labels are shown in black while error examples are shown in red. tures from fc7 layer of Resnet50 [24] to 192-d. • AdjMat-only directly feeds the explicitly learned adja￾cency matrix …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 61 canonical work pages

  1. [1]

    Data-driven model of nonverbal behavior for socially assistive human-robot inter- actions

    Henny Admoni and Brian Scassellati. Data-driven model of nonverbal behavior for socially assistive human-robot inter- actions. In ICMI, 2014. 3

    work page 2014

  2. [2]

    Social eye gaze in human-robot interaction: A review

    Henny Admoni and Brian Scassellati. Social eye gaze in human-robot interaction: A review. JHRI, 6(1), 2017. 1, 2, 3

    work page 2017

  3. [3]

    Look like me: matching robot personality via gaze to increase motiva- tion

    Sean Andrist, Bilge Mutlu, and Adriana Tapus. Look like me: matching robot personality via gaze to increase motiva- tion. In CHI, 2015. 3

    work page 2015

  4. [4]

    Conversational gaze aversion for humanlike robots

    Sean Andrist, Xiang Zhi Tan, Michael Gleicher, and Bilge Mutlu. Conversational gaze aversion for humanlike robots. In HRI, 2014. 3

    work page 2014

  5. [5]

    Gaze and mutual gaze

    Michael Argyle and Mark Cook. Gaze and mutual gaze. Cambridge U Press, 1976. 2

    work page 1976

  6. [6]

    Battaglia, Razvan Pascanu, Matthew Lai, Danilo Rezende, and Koray Kavukcuoglu

    Peter W. Battaglia, Razvan Pascanu, Matthew Lai, Danilo Rezende, and Koray Kavukcuoglu. Interaction networks for learning about objects, relations and physics. In NIPS, 2016. 3

    work page 2016

  7. [7]

    State-of-the-art in visual attention modeling

    Ali Borji and Laurent Itti. State-of-the-art in visual attention modeling. IEEE TPAMI, 35(1):185–207, 2013. 3

    work page 2013

  8. [8]

    Dominey, and Jocelyne Ventre-Dominey

    Jean-David Boucher, Ugo Pattacini, Amelie Lelong, Gerard Bailly, Frederic Elisei, Sascha Fagel, Peter F. Dominey, and Jocelyne Ventre-Dominey. I reach faster when i see you look: gaze effects in human-human and human-robot face-to-face cooperation. Frontiers in Neurorobotics, 6:3, 2012. 3

    work page 2012

Show all 64 references
  1. [9]

    Meltzoff

    Rechele Brooks and Andrew N. Meltzoff. The development of gaze following and its relation to language. Developmen- tal Science, 8(6):535–543, 2005. 2

    2005

  2. [10]

    Spectral networks and locally connected networks on graphs

    Joan Bruna, Wojciech Zaremba, Arthur Szlam, and Yann Le- cun. Spectral networks and locally connected networks on graphs. In ICLR, 2014. 3

    2014

  3. [11]

    Burgoon, Laura K

    Judee K. Burgoon, Laura K. Guerrero, and Kory Floyd. Non- verbal communication. Routledge, 2016. 3

    2016

  4. [12]

    Iter- ative visual reasoning beyond convolutions

    Xinlei Chen, Li-Jia Li, Li Fei-Fei, and Abhinav Gupta. Iter- ative visual reasoning beyond convolutions. In CVPR, 2018. 3

    2018

  5. [13]

    Discriminative embed- dings of latent variable models for structured data

    Hanjun Dai, Bo Dai, and Le Song. Discriminative embed- dings of latent variable models for structured data. In ICML,

  6. [14]

    Duvenaud, Dougal Maclaurin, Jorge Iparraguirre, Rafael Bombarell, Timothy Hirzel, Al´an Aspuru-Guzik, and Ryan P

    David K. Duvenaud, Dougal Maclaurin, Jorge Iparraguirre, Rafael Bombarell, Timothy Hirzel, Al´an Aspuru-Guzik, and Ryan P. Adams. Convolutional networks on graphs for learn- ing molecular fingerprints. In NIPS, 2015. 3

    2015

  7. [15]

    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

  8. [16]

    Shifting more attention to video salient object detection

    Deng-Ping Fan, Wenguan Wang, Ming-Ming Cheng, and Jianbing Shen. Shifting more attention to video salient object detection. In CVPR, 2019. 3

    2019

  9. [17]

    Inferring shared attention in social scene videos

    Lifeng Fan, Yixin Chen, Ping Wei, Wenguan Wang, and Song-Chun Zhu. Inferring shared attention in social scene videos. In CVPR, 2018. 3

    2018

  10. [18]

    Pairwise body-part attention for recognizing human-object interactions

    Hao-Shu Fang, Jinkun Cao, Yu-Wing Tai, and Cewu Lu. Pairwise body-part attention for recognizing human-object interactions. In ECCV, 2018. 3

    2018

  11. [19]

    Joint 3d face reconstruction and dense alignment with position map regression network

    Yao Feng, Fan Wu, Xiaohu Shao, Yanfeng Wang, and Xi Zhou. Joint 3d face reconstruction and dense alignment with position map regression network. In ECCV, 2018. 8

    2018

  12. [20]

    Schoenholz, Patrick F

    Justin Gilmer, Samuel S. Schoenholz, Patrick F. Riley, Oriol Vinyals, and George E. Dahl. Neural message passing for quantum chemistry. In ICML, 2017. 3, 5

    2017

  13. [21]

    Glenberg, Jennifer L

    Arthur M. Glenberg, Jennifer L. Schroeder, and David A. Robertson. Averting the gaze disengages the environ- ment and facilitates remembering. Memory & Cognition , 26(4):651–658, 1998. 2

    1998

  14. [22]

    Haith, Terry Bergman, and Michael J

    Marshall M. Haith, Terry Bergman, and Michael J. Moore. Eye contact and face scanning in early infancy. Science, 198(4319):853–855, 1977. 2

    1977

  15. [23]

    Inductive representation learning on large graphs

    Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. In NIPS, 2017. 3

    2017

  16. [24]

    Deep residual learning for image recognition

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

  17. [25]

    Chien-Ming Huang and Andrea L. Thomaz. Effects of re- sponding to, initiating and ensuring joint attention in human- robot interaction. In 2011 Ro-Man, 2011. 3

    2011

  18. [26]

    Itier and Magali Batty

    Roxane J. Itier and Magali Batty. Neural bases of eye and gaze processing: the core of social cognition. Neuroscience & Biobehavioral Reviews, 33(6):843–863, 2009. 2

    2009

  19. [27]

    A model of saliency-based visual attention for rapid scene analysis

    Laurent Itti, Christof Koch, and Ernst Niebur. A model of saliency-based visual attention for rapid scene analysis. IEEE TPAMI, 20(11):1254–1259, 1998. 3

    1998

  20. [28]

    Zamir, Silvio Savarese, and Ashutosh Saxena

    Ashesh Jain, Amir R. Zamir, Silvio Savarese, and Ashutosh Saxena. Structural-RNN: Deep learning on spatio-temporal graphs. In CVPR, 2016. 3

    2016

  21. [29]

    social gaze space

    Mathis Jording, Arne Hartz, Gary Bente, Martin Schulte- R¨uther, and Kai V ogeley. The “social gaze space”: A tax- onomy for gaze-based communication in triadic interactions. Frontiers in Psychology, 9:226, 2018. 2

    2018

  22. [30]

    Neural relational inference for interacting systems

    Thomas Kipf, Ethan Fetaya, Kuan-Chieh Wang, Max Welling, and Richard Zemel. Neural relational inference for interacting systems. In ICML, 2018. 3, 5

    2018

  23. [31]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. Semi-supervised classi- fication with graph convolutional networks. In ICLR, 2017. 3

    2017

  24. [32]

    Chris L. Kleinke. Gaze and eye contact: a research review. Psychological Bulletin, 100(1):78, 1986. 3

    1986

  25. [33]

    Unique morphology of the human eye

    Hiromi Kobayashi and Shiro Kohshima. Unique morphology of the human eye. Nature, 387(6635):767, 1997. 2

    1997

  26. [34]

    Situation recognition with graph neural networks

    Ruiyu Li, Makarand Tapaswi, Renjie Liao, Jiaya Jia, Raquel Urtasun, and Sanja Fidler. Situation recognition with graph neural networks. In ICCV, 2017. 3

    2017

  27. [35]

    Gated graph sequence neural networks

    Yujia Li, Daniel Tarlow, Marc Brockschmidt, and Richard Zemel. Gated graph sequence neural networks. In ICML,

  28. [36]

    Ishi, Hiroshi Ishiguro, and Norihiro Hagita

    Chaoran Liu, Carlos T. Ishi, Hiroshi Ishiguro, and Norihiro Hagita. Generation of nodding, head tilting and eye gazing for human-robot dialogue interaction. In HRI, 2012. 3

    2012

  29. [37]

    The more you know: Using knowledge graphs for image classification

    Kenneth Marino, Ruslan Salakhutdinov, and Abhinav Gupta. The more you know: Using knowledge graphs for image classification. In CVPR, 2017. 3

    2017

  30. [38]

    Meltzoff, Rechele Brooks, Aaron P

    Andrew N. Meltzoff, Rechele Brooks, Aaron P. Shon, and Rajesh P.N. Rao. social robots are psychological agents for infants: A test of gaze following. Neural networks, 23(8- 9):966–972, 2010. 3

    2010

  31. [39]

    Dunham, and Phil Dunham

    Chris Moore, Philip J. Dunham, and Phil Dunham. Joint attention: Its origins and role in development . Psychology Press, 2014. 2

    2014

  32. [40]

    A story- telling robot: Modeling and evaluation of human-like gaze behavior

    Bilge Mutlu, Jodi Forlizzi, and Jessica Hodgins. A story- telling robot: Modeling and evaluation of human-like gaze behavior. In IEEE-RAS ICHR, 2006. 3

    2006

  33. [41]

    Conversational gaze mechanisms for humanlike robots

    Bilge Mutlu, Takayuki Kanda, Jodi Forlizzi, Jessica Hod- gins, and Hiroshi Ishiguro. Conversational gaze mechanisms for humanlike robots. ACM TIIS, 1(2):12, 2012. 3

    2012

  34. [42]

    Learning convolutional neural networks for graphs

    Mathias Niepert, Mohamed Ahmed, and Konstantin Kutzkov. Learning convolutional neural networks for graphs. In ICML, 2016. 3

    2016

  35. [43]

    Development of joint attention in infant chimpanzees

    Sanae Okamoto-Barth and Masaki Tomonaga. Development of joint attention in infant chimpanzees. In Cognitive devel- opment in chimpanzees, pages 155–171. Springer, 2006. 2

    2006

  36. [44]

    Park, Eakta Jain, and Yaser Sheikh

    Hyun S. Park, Eakta Jain, and Yaser Sheikh. 3d social saliency from head-mounted cameras. In NIPS, 2012. 3

    2012

  37. [45]

    Social saliency prediction

    Hyun Soo Park and Jianbo Shi. Social saliency prediction. In CVPR, 2015. 3

    2015

  38. [46]

    Learning human-object interactions by graph parsing neural networks

    Siyuan Qi, Wenguan Wang, Baoxiong Jia, Jianbing Shen, and Song-Chun Zhu. Learning human-object interactions by graph parsing neural networks. In ECCV, 2018. 2, 3, 5

    2018

  39. [47]

    Morris D. Riemer. The averted gaze. Psychiatric Quarterly, 23(1):108–115, 1949. 2

    1949

  40. [48]

    Barrett, Mateusz Malinowski, Razvan Pascanu, Peter Battaglia, and Tim Lilli- crap

    Adam Santoro, David Raposo, David G. Barrett, Mateusz Malinowski, Razvan Pascanu, Peter Battaglia, and Tim Lilli- crap. A simple neural network module for relational reason- ing. In NIPS, 2017. 3

    2017

  41. [49]

    The graph neural net- work model

    Franco Scarselli, Marco Gori, Ah Chung Tsoi, Markus Ha- genbuchner, and Gabriele Monfardini. The graph neural net- work model. IEEE TNNLS, 20(1):61–80, 2009. 3

    2009

  42. [50]

    Johnson, and Gergely Csibra

    Atsushi Senju, Mark H. Johnson, and Gergely Csibra. The development and neural basis of referential gaze perception. Social neuroscience, 1(3-4):220–234, 2006. 2

    2006

  43. [51]

    Following gaze: Gaze-following behav- ior as a window into social cognition.Frontiers in integrative neuroscience, 4:5, 2010

    Stephen Shepherd. Following gaze: Gaze-following behav- ior as a window into social cognition.Frontiers in integrative neuroscience, 4:5, 2010. 2

    2010

  44. [52]

    Dynamic edge- conditioned filters in convolutional neural networks on graphs

    Martin Simonovsky and Nikos Komodakis. Dynamic edge- conditioned filters in convolutional neural networks on graphs. In CVPR, 2017. 3

    2017

  45. [53]

    Very deep convo- lutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep convo- lutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014. 8

    2014 arXiv

  46. [54]

    Vasant Srinivasan and Robin R. Murphy. A survey of social gaze. In HRI, 2011. 3

    2011

  47. [55]

    Maria Staudte and Matthew W. Crocker. Visual attention in spoken human-robot interaction. In HRI, 2009. 3

    2009

  48. [56]

    Maria Staudte and Matthew W. Crocker. Investigating joint attention mechanisms through spoken human-robot interac- tion. Cognition, 120(2):268–291, 2011. 3

    2011

  49. [57]

    Learning multiagent communication with backpropagation

    Sainbayar Sukhbaatar, Arthur Szlam, and Rob Fergus. Learning multiagent communication with backpropagation. In NIPS, 2016. 3

    2016

  50. [58]

    Mataric, and Brian Scassellati

    Adriana Tapus, Maja J. Mataric, and Brian Scassellati. The grand challenges in socially assistive robotics. IEEE Robotics & Automation Magazine, 14(1):35–42, 2007. 3

    2007

  51. [59]

    Origins of human communication

    Michael Tomasello. Origins of human communication. MIT Press, 2010. 1

    2010

  52. [60]

    Graph at- tention networks

    Petar Velickovic, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. Graph at- tention networks. In ICLR, 2018. 2, 3, 5

    2018

  53. [61]

    Ef- ficiently scaling up crowdsourced video annotation

    Carl V ondrick, Donald Patterson, and Deva Ramanan. Ef- ficiently scaling up crowdsourced video annotation. IJCV, 101(1):184–204, 2013. 4

    2013

  54. [62]

    Salient object detection in the deep learning era: An in-depth survey

    Wenguan Wang, Qiuxia Lai, Huazhu Fu, Jianbing Shen, and Haibin Ling. Salient object detection in the deep learning era: An in-depth survey. arXiv preprint arXiv:1904.09146,

    1904 arXiv

  55. [63]

    Deep visual attention prediction

    Wenguan Wang and Jianbing Shen. Deep visual attention prediction. IEEE TIP, 27(5):2368–2378, 2018. 3

    2018

  56. [64]

    Gaze following

    Klaus Zuberb ¨uhler. Gaze following. Current Biology , 18(11):R453–R455, 2008. 2

    2008

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.