Pith. sign in

REVIEW 4 major objections 8 minor 59 references

ASTRA: A Scene-aware TRAnsformer-based model for trajectory prediction

T0 review · 4 major / 8 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read ASTRA, a lightweight scene-aware transformer, reports average gains of 27% deterministic and 10% stochastic on ETH-UCY, 26% on PIE, with seven times fewer parameters than the leading comparison model.

desk verdict A lightweight transformer with strong benchmark numbers, but the headline gains rest partly on an undisclosed, benchmark-selected loss weighting. read the letter →

arxiv 2501.09878 v1 pith:VJTVM4FY submitted 2025-01-16 cs.CV cs.AI

Izzeddin Teeti , Aniket Thomas , Munish Monga , Sachin Kumar , Uddeshya Singh , Andrew Bradley , Biplab Banerjee , Fabio Cuzzolin This is my paper · ORCID
classification cs.CVcs.AI
keywords pedestriantrajectorypredictiongraph-awaretransformerRandomWalkPositionalEncodingU-NetscenefeaturesconditionalvariationalautoencoderweightedpenaltylossETH-UCYPIEdataset
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 ASTRA, a lightweight pedestrian trajectory prediction model that fuses scene context, spatial dynamics, social interactions, and temporal progression in a single transformer-based architecture. The central claim is that this design, together with a parabolic weighted penalty loss, improves average prediction error by 27% in the deterministic setting and 10% in the stochastic setting on ETH-UCY, and by 26% on PIE, while using roughly seven times fewer parameters than the leading comparison model. If true, the result matters because autonomous vehicles and robots need accurate trajectory forecasts that fit on limited onboard hardware, and this model claims to improve both accuracy and efficiency at once. The paper further claims that the same model works from both bird's-eye-view and ego-vehicle-view cameras, a versatility most prior trajectory predictors lack.

What carries the argument

The central mechanism is the joint agent-scene embedding. Spatial coordinates are projected by an MLP, sinusoidal temporal encodings are added, and social encodings come from a fully connected undirected graph with edge weights $e_{ij} = 1/d(v_i, v_j)$, whose structure is captured by Random Walk Positional Encodings; these three streams are concatenated and passed through a single-layer agent-aware transformer encoder. In parallel, a U-Net encoder pre-trained with the Weighted Hausdorff Distance produces scene tokens that go through a single-layer scene-aware transformer encoder. The second key piece is the weighted penalty loss $w(t) = (\alpha-\beta)\left(2t/T_{\mathrm{pred}} - 1\right)^2 + \beta$, which assigns maximum weight to the first and last predicted timesteps and minimum weight at the midpoint; the paper reports that this parabolic form beats linear and quadratic alternatives on the UNIV fold.

What would settle it

Re-run ASTRA on all five ETH-UCY scenes and PIE with $\alpha$ and $\beta$ fixed before any test-set comparison, and with the penalty form fixed to the same function used in prior work; if the 27%/10%/26% margins do not survive, the central claim fails. A simpler check: train the same architecture with the unpenalised loss and with each of the three penalty forms, then report per-scene test results for all ETH-UCY scenes rather than only UNIV, to see whether the parabolic advantage is consistent.

Watch

Extended reading notes

Core claim

On the paper's own terms, ASTRA establishes that encoding the scene with a frozen U-Net latent representation and feeding those tokens through a single-layer transformer, while the agent stream concatenates spatial, temporal, and social encodings in a graph-aware transformer, yields a trajectory predictor that beats published deterministic and stochastic baselines on ETH-UCY and on PIE. The social graph is built from inverse pairwise distances and represented with Random Walk Positional Encodings, so the transformer can attend over agents while preserving their structural relations. The stochastic variant adds a conditional variational autoencoder that samples K future trajectories, and the training objective is a weighted loss that up-weights errors at the start and end of the prediction horizon. The paper's claimed novelty is the integration itself: spatial, temporal, and social dimensions are processed in one attention pass, rather than by separate GNN and transformer blocks, and the scene encoder captures both labeled and unlabeled pedestrians.

Load-bearing premise

The reported gains depend on the parabolic penalty function $w(t)$, whose parameters $\alpha$ and $\beta$ are never disclosed and whose functional form was selected after comparing linear, quadratic, and parabolic variants on the UNIV fold; if this choice amounts to tuning on the evaluation metric, the headline percentages partly reduce to fitted values.

Editorial extensions

If this is right

  • ASTRA reports average deterministic ADE/FDE of 0.38/0.70 on ETH-UCY, beating EqMotion's 0.49/1.03, and stochastic minADE20/minFDE20 of 0.20/0.28, beating LeapFrog's 0.21/0.33.
  • The full architecture uses 1.56M parameters and 1.77M FLOPs, versus 6.78M parameters and 3.08G FLOPs for AgentFormer; a deterministic variant without the U-Net can be as small as 13.52K parameters.
  • Adding frame encodings from the U-Net improves the stochastic ETH-UCY average from 0.27/0.38 to 0.20/0.28, so scene context is a direct source of the reported gain.
  • The parabolic penalty improves stochastic UNIV minADE20/minFDE20 from 0.37/0.52 unpenalised to 0.28/0.41, and the paper shows similar qualitative gains for penalised versus unpenalised trajectories.
  • Because the same architecture is evaluated on BEV (ETH-UCY) and EVV (PIE) inputs, the paper claims a single model can serve both ego-vehicle and overhead camera settings.

Reading between the lines

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

  • A testable extension not in the paper: fix $\alpha$ and $\beta$ on one dataset and apply them unchanged to another; if the parabolic penalty is a genuine inductive bias, the gains should transfer without per-dataset tuning.
  • The scene encoder is a frozen U-Net, which suggests the scene branch could be swapped for any pretrained latent scene representation; ablating with a randomly initialized U-Net would show how much of the gain comes from keypoint pretraining versus architecture.
  • The 13.52K-parameter deterministic variant points toward embedded deployment, but the paper reports FLOPs and parameter counts, not latency; measuring inference time on a single GPU or edge device would complete the efficiency picture.
  • The comparison set for the 'seven times fewer parameters' claim is the leading stochastic model; a uniform parameter table for every baseline in Tables 1 and 2 would make the efficiency advantage directly checkable.
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 / 8 minor

Summary. ASTRA is a lightweight pedestrian trajectory prediction model combining three components: a frozen U-Net keypoint extractor that encodes scene context without segmentation supervision, a single-layer agent-aware transformer whose tokens carry random-walk positional encodings of a fully connected inter-agent graph, and a CVAE branch for multimodal predictions. The methodological addition is a weighted penalty loss, w(t) = (α−β)*(2t/Tpred − 1)^2 + β (Appendix C.3), applied to per-timestep MSE or Smooth-L1 terms. On ETH-UCY the paper reports average ADE/FDE of 0.38/0.70 in the deterministic setting (a 27% improvement over EqMotion) and minADE20/minFDE20 of 0.20/0.28 in the stochastic setting (10% over LeapFrog), together with a 26% improvement over PedFormer on PIE and a seven-fold reduction in parameters relative to LeapFrog. Ablations (Tables 2, 3, 5, 6) trace the contributions of each component and of the penalty. The central claim is that ASTRA is the most accurate lightweight pedestrian trajectory predictor on these benchmarks while remaining deployable on resource-constrained devices.

Significance. If the reported numbers survive a properly controlled evaluation, this is a solid engineering contribution to lightweight multi-agent trajectory prediction: the RWPE-based graph-aware transformer cleanly injects inter-agent structure into attention, the frozen U-Net scene encoder avoids segmentation-map supervision, and the model is genuinely small. Internal consistency is a strength: the component ablation (Table 3) is monotone and matches the design narrative, and the loss ablations (Tables 2, 5, 6) consistently show a benefit from the penalty. The significance is nonetheless capped by three verification gaps: the stochastic SOTA claim rests entirely on the weighted loss (the unpenalized model, 0.27/0.38 in Table 2, is worse than LeapFrog) while the loss parameters are undisclosed and the parabolic form was chosen on the UNIV test fold; the 26% PIE headline uses an unspecified aggregation that excludes the ARB regression; and no seeds or error bars are reported, so a minADE difference of 0.01 cannot be assessed. The paper ships no code or models; all claims rest on tabulated point estimates.

major comments (4)
  1. [§3.1.7, Appendix C.3, Tables 2 and 6] The weighted penalty loss is load-bearing in a stronger sense than the paper acknowledges. In Table 2, the unpenalized ASTRA (0.27/0.38 average minADE20/minFDE20) is worse than LeapFrog (0.21/0.33), so the entire stochastic improvement of 10% over the SOTA baseline is attributable to the weighted-loss component rather than to the architecture. However, the two parameters of the parabolic weight w(t) = (α−β)*(2t/Tpred − 1)^2 + β (Appendix C.3, Eq. 20) are never reported, and the parabolic form was selected by comparing linear, quadratic, and parabolic variants on the UNIV test fold (Table 6), the same UNIV fold whose stochastic results (0.28/0.41 in Table 2) enter the headline average. This is a correctness risk: the reported gain may be partly fitted to the evaluation fold. Please report α and β for every configuration, describe the selection protocol, and demonstrate that the parabolic form and parameters transfer to the remaining folds, for example by fixing the form and α, β a priori on a validation split and then reporting the other four folds' results.
  2. [Abstract and §4.5.1, Table 4] The '26% improvement on PIE' headline is not reproducible from Table 4 under a standard aggregation. Relative to PedFormer, ASTRA improves CADE by 24.2% (13.08->9.91), CFDE by 26.1% (30.35->22.42), and FRB by 47.9% (32.79->17.07), but ARB worsens by 20.0% (15.27->18.32). The mean of the four per-metric relative changes is about 19.6%, while the mean over only the improved metrics is about 32.8%; the stated 26% must therefore use an undefined or selective aggregation. Please state the aggregation formula explicitly, and present the ARB regression in the results discussion rather than only in the table, since it materially qualifies the claim that ASTRA outperforms PedFormer.
  3. [§4.3, Tables 1–4] All reported results are single point estimates; no seeds, repeated runs, or variance statistics are given. The stochastic headline is an improvement of 0.01 in minADE (0.21->0.20) and 0.05 in minFDE (0.33->0.28) over LeapFrog (Table 2), and four baselines in the same table sit at minADE 0.21. Seed-to-seed variation in this benchmark family is typically of the same order as these deltas, so without error bars the claimed 10% improvement is not statistically assessable. Please report mean±std over at least five seeds for all ASTRA configurations, and for the principal baselines where the underlying runs are available.
  4. [Abstract, §4.4, §4.5.1, Figure 1] The efficiency claim is internally inconsistent. The abstract and §4.5.1 say ASTRA has 'seven times fewer trainable parameters' than LeapFrog, but the same section states that the reported count 'includes the parameters from the U-Net, which is otherwise actually frozen'; frozen weights are not trainable, so the comparison basis is unclear, and the LeapFrog parameter count is never stated. Section 4.4 gives three different FLOPs figures for ASTRA (1.7 MFLOPs, 839 KFLOPs, 16 KFLOPs) without specifying the configuration or the per-input unit (per agent, per sequence, or per frame), and the comparison with AgentFormer (3.084 GFLOPs) mixes units across models. Please adopt a single counting convention (trainable vs total, with or without the frozen U-Net), state the LeapFrog parameter count, and compute all FLOPs under one stated convention.
minor comments (8)
  1. [§2.4 and §3.1.4] The claim that ASTRA is 'the first work towards utilising a graph-aware transformer to solve the trajectory prediction problem' is not substantiated against works already in the reference list (e.g., STAR [48], Social Graph Transformer Networks [9]); please temper the priority claim or replace it with a direct comparison.
  2. [§3.1.7] Equations (11) and (12) state the same final loss twice, once in generative-model notation and once in Gaussian notation; one of the two should be deleted.
  3. [§3.1.4] The sentence 'temporal encoding (equation in Supplementary material) is also included to distinguish them' points to a supplementary equation, but the temporal encoding is already defined in Eq. (1) of the main text; the cross-reference should be fixed.
  4. [Appendix A] The sentence 'the model pushes the learned distribution pθ(zp|x) to be as close as possible to the ground truth distribution qΦ(zq|x,y)' reverses the standard CVAE roles, in which the KL term drives the posterior toward the prior; please rephrase.
  5. [Abstract and §4.5.1] The aggregation behind 'average improvement of 27%/10%' should be stated once in the paper (e.g., for the deterministic case, the mean of the per-metric relative improvements (0.49−0.38)/0.49 and (1.03−0.70)/1.03), so that the headline is reproducible from Tables 1 and 2.
  6. [§3.1.2 and §4.1] The U-Net keypoint extractor pretraining is described only by reference to [13]; please state which frames and labels (ETH-UCY, PIE, or other data) were used, since the scene-aware component depends on this step.
  7. [Throughout] Notation is inconsistent ('CV AE' vs 'CVAE', 'sinosuidal' in Figure 3, mid-sentence capitalizations such as 'During Training'), and the paper would benefit from a copy-editing pass.
  8. [§4.3] Please report the number of stochastic samples K used at inference and the augmentation parameters, as these directly affect the minADE20/minFDE20 comparisons.

Circularity Check

1 steps flagged · score 6.0 of 10

The headline accuracy gain is partly a product of selecting the weighted-penalty loss on the UNIV evaluation fold: the parabolic loss form is chosen via Table 6 on the same split whose results are then included in the claimed 10% stochastic improvement.

  1. fitted input called prediction [Section 3.1.7, Appendix C.3, Table 6, and Table 2]
    "To address this issue, we experimented with a parabolic weighting function for the penalty. Empirically, this approach outperformed the linear and quadratic strategies... w(t) = (α − β) · (2 · t/Tpred − 1)^2 + β ... Table 6: Ablation: Comparing penalization strategies with SL1 loss on ETH-UCY (UNIV) dataset... Unpenalised 0.37/0.52; Linear 0.33/0.47; Quadratic 0.30/0.46; Parabolic 0.28/0.41"

    The stochastic results that support the headline 10% improvement are produced with a loss whose functional form was chosen by ranking unpenalised/linear/quadratic/parabolic variants on the UNIV split of ETH-UCY (Table 6). The same split is then included in the average stochastic numbers (Table 2: UNIV 0.28/0.41, average 0.20/0.28) compared with LeapFrog. The parabolic penalty is therefore not an independently derived prediction mechanism; its reported benefit over the alternatives is by construction the best among the compared choices on that fold. α and β are never disclosed, so the gain cannot be re-checked as a parameter-free out-of-sample result. This is benchmark selection called a model contribution, though the architecture itself is independently testable.

full rationale

ASTRA's architectural derivation is largely self-contained: the U-Net scene encoder, RWPE-based graph-aware transformer, and CVAE are assembled from externally cited components, and the parameter-count claims are mechanically checkable. The circular component is the weighted-penalty loss, which is the main driver of the stochastic improvement. The parabolic weight function in Eq. 20 was selected by comparing penalty variants on the UNIV fold (Table 6), and the same fold contributes to the average results used in the abstract's 10% stochastic improvement claim. This is benchmark overfitting rather than a definitional equivalence: the model still learns from data, so the circularity is partial (score 6). If the loss parameters and selection procedure had been validated on held-out data, this would be ordinary hyperparameter tuning. The PIE claim is also selective, since ARB worsens from 15.27 (PedFormer) to 18.32 (ASTRA) while CADE/CFDE/FRB improve, but that is a reporting concern, not a circular derivation. No load-bearing self-citations or imported uniqueness theorems appear in the paper, so no additional circularity steps are warranted.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

All free parameters relate to the loss and architecture; no new physical entities are postulated. The loss parameters are the most consequential because they are selected based on the target metric.

free parameters (2)
  • Parabolic penalty parameters alpha and beta = not disclosed
    The weight function w(t) = (alpha-beta)*(2*t/T_pred - 1)^2 + beta is introduced in Sec. 3.1.7 and Appendix C.3, but the values of alpha and beta are never reported; the parabolic form was chosen after comparing linear, quadratic and parabolic variants on the UNIV dataset (Table 6), so these are hand-tuned to the benchmark.
  • Architecture hyperparameters (transformer width, number of layers, MLP dimensions) = not reported
    The performance depends on these, and without them reproduction is impossible; some may have been selected on validation performance, though the paper does not describe the selection process.
assumptions (3)
  • domain assumption Random Walk Positional Encodings (RWPE) provide informative structural encodings for the agent graph.
    Taken from Dwivedi et al. (2021); the paper relies on this to make the transformer graph-aware (Section 3.1.4).
  • domain assumption The U-Net encoder pretrained with Weighted Hausdorff Distance yields a scene representation useful for trajectory prediction.
    The latent vector from the U-Net is used as scene context (Section 3.1.2), but the pretraining is from Ribera et al. (2019) and is not re-evaluated here.
  • domain assumption Evaluating with a single random seed without error bars is sufficient to establish relative performance.
    All reported metrics are single numbers; the paper implicitly assumes the differences are not noise (Tables 1-4).

how reviews work

0 comments
Cite this review

Pith. "Pith review of ASTRA: A Scene-aware TRAnsformer-based model for trajectory prediction." pith.science (2026). https://pith.science/paper/VJTVM4FY

@misc{pith2026250109878,
  author       = {Pith},
  title        = {Pith review of: ASTRA: A Scene-aware TRAnsformer-based model for trajectory prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VJTVM4FY}},
  note         = {Machine review of arXiv:2501.09878}
}
read the original abstract

We present ASTRA (A} Scene-aware TRAnsformer-based model for trajectory prediction), a light-weight pedestrian trajectory forecasting model that integrates the scene context, spatial dynamics, social inter-agent interactions and temporal progressions for precise forecasting. We utilised a U-Net-based feature extractor, via its latent vector representation, to capture scene representations and a graph-aware transformer encoder for capturing social interactions. These components are integrated to learn an agent-scene aware embedding, enabling the model to learn spatial dynamics and forecast the future trajectory of pedestrians. The model is designed to produce both deterministic and stochastic outcomes, with the stochastic predictions being generated by incorporating a Conditional Variational Auto-Encoder (CVAE). ASTRA also proposes a simple yet effective weighted penalty loss function, which helps to yield predictions that outperform a wide array of state-of-the-art deterministic and generative models. ASTRA demonstrates an average improvement of 27%/10% in deterministic/stochastic settings on the ETH-UCY dataset, and 26% improvement on the PIE dataset, respectively, along with seven times fewer parameters than the existing state-of-the-art model (see Figure 1). Additionally, the model's versatility allows it to generalize across different perspectives, such as Bird's Eye View (BEV) and Ego-Vehicle View (EVV).

Figures

Figures reproduced from arXiv: 2501.09878 by the authors.

Figure 1
Figure 1. Comparison of average (minADE20/minFDE20) against the number of parameters for various models on the ETH-UCY dataset. Each point represents a different model, with the model name and number of parameters in millions indicated. Our model, ASTRA, achieves state-of-the-art results with the least number of parameters, demonstrating its efficiency and effectiveness in pedestrian trajectory forecasting. 1 Introduction The… view at source ↗
Figure 2
Figure 2. Sample images from the BEV dataset (ETH), and EVV dataset (PIE), along with their Grad-CAM [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Model Architecture. Overview of ASTRA model architecture for pedestrian trajectory forecasting. details, the latter focuses on encoding the spatial, temporal, and social dimensions of the agents, as shown in [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Pretraining U-Net based keypoint extractor. A latent representation of pedestrian characteristics is obtained using a pre-trained U-Net encoder ( [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Grad-CAM visualizations: In (a), the red circle indicates unlabelled pedestrians, while the green square highlights labelled pedestrians. In (b), the U-Net-based keypoint extractor focuses on unlabelled pedestrians as well, thereby capturing scene context from them too…
Figure 6
Figure 6. Figure 6: Sample images of the deterministic prediction from BEV datasets (a.) (ETH and UCY) and [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Visual comparison of penalised and unpenalised loss on ETH-UCY, showing the en￾hanced performance of the former [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 9
Figure 9. Figure 9: Comparison of various weighted penalty strategies [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]
Figure 10
Figure 10. Figure 10: Qualitative comparison of unpenalised vs. penalised trajectories on ETH-UCY dataset in stochas [PITH_FULL_IMAGE:figures/full_fig_p018_10.png]
Figure 11
Figure 11. Figure 11: Qualitative comparison of unpenalised vs. penalised trajectories on ETH-UCY dataset in stochas [PITH_FULL_IMAGE:figures/full_fig_p018_11.png]
Figure 12
Figure 12. Figure 12: Multi-modal trajectory visualizations on ETH-UCY dataset (BEV) [PITH_FULL_IMAGE:figures/full_fig_p020_12.png]
Figure 13
Figure 13. Figure 13: Deterministic trajectory visualizations on ETH-UCY dataset (BEV) [PITH_FULL_IMAGE:figures/full_fig_p021_13.png]
Figure 14
Figure 14. Figure 14: Trajectory Visualizations on PIE Dataset (EVV) where the red and cyan bounding box indicates [PITH_FULL_IMAGE:figures/full_fig_p022_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 46 canonical work pages

  1. [2]

    Pedformer: Pedestrian behavior prediction via cross-modal attention modulation and gated multitask learning,

    A. Rasouli and I. Kotseruba, “Pedformer: Pedestrian behavior prediction via cross-modal attention modulation and gated multitask learning,” in 2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023, pp. 9844–9851

    work page 2023

  2. [3]

    Trajectory prediction with latent belief energy-based model,

    B. Pang, T. Zhao, X. Xie, and Y. N. Wu, “Trajectory prediction with latent belief energy-based model,” in 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 11 809– 11 819

    work page 2021

  3. [4]

    Semi-supervised classification with graph convolutional networks,

    T. N. Kipf and M. Welling, “Semi-supervised classification with graph convolutional networks,” CoRR, vol. abs/1609.02907, 2016. [Online]. Available: http://arxiv.org/abs/1609.02907

    arXiv 2016

  4. [5]

    Eqmotion: Equivariant multi-agent motion prediction with invariant interaction reasoning,

    C. Xu, R. T. Tan, Y. Tan, S. Chen, Y. G. Wang, X. Wang, and Y. Wang, “Eqmotion: Equivariant multi-agent motion prediction with invariant interaction reasoning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 1410–1420

    work page 2023

  5. [6]

    Graph-based spatial transformer with memory replay for multi- future pedestrian trajectory prediction,

    L. Li, M. Pagnucco, and Y. Song, “Graph-based spatial transformer with memory replay for multi- future pedestrian trajectory prediction,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 2231–2241

    work page 2022

  6. [7]

    Vnagt: Variational non-autoregressive graph transformer network for multi-agent trajectory prediction,

    X. Chen, H. Zhang, Y. Hu, J. Liang, and H. Wang, “Vnagt: Variational non-autoregressive graph transformer network for multi-agent trajectory prediction,”IEEE Transactions on Vehicular Technology, pp. 1–12, 2023

    work page 2023

  7. [8]

    Hdgt: Heterogeneous driving graph transformer for multi-agent trajectory prediction via scene encoding,

    X. Jia, P. Wu, L. Chen, Y. Liu, H. Li, and J. Yan, “Hdgt: Heterogeneous driving graph transformer for multi-agent trajectory prediction via scene encoding,” IEEE transactions on pattern analysis and machine intelligence, 2023

    work page 2023

  8. [9]

    Social graph transformer networks for pedestrian trajectory prediction in complex social scenarios,

    Y. Liu, L. Yao, B. Li, X. Wang, and C. Sammut, “Social graph transformer networks for pedestrian trajectory prediction in complex social scenarios,” in Proceedings of the 31st ACM International Conference on Information & Knowledge Management, ser. CIKM ’22. New York, NY, USA: Association for Computing Machinery, 2022, p. 1339–1349. [Online]. Available: h...

    arXiv 2022

Show all 59 references
  1. [10]

    Bifold and semantic reasoning for pedestrian behavior prediction,

    A. Rasouli, M. Rohani, and J. Luo, “Bifold and semantic reasoning for pedestrian behavior prediction,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 15 600–15 610

    2021

  2. [11]

    From goals, waypoints & paths to long term human trajectory forecasting,

    K. Mangalam, Y. An, H. Girase, and J. Malik, “From goals, waypoints & paths to long term human trajectory forecasting,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 15 233–15 242

    2021

  3. [12]

    Bitrap: Bi-directional pedestrian trajectory prediction with multi-modal goal estimation,

    Y. Yao, E. Atkins, M. Johnson-Roberson, R. Vasudevan, and X. Du, “Bitrap: Bi-directional pedestrian trajectory prediction with multi-modal goal estimation,” IEEE Robotics and Automation Letters, vol. 6, no. 2, pp. 1463–1470, 2021

    2021

  4. [13]

    Locating objects without bounding boxes,

    J. Ribera, D. Guera, Y. Chen, and E. J. Delp, “Locating objects without bounding boxes,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 6479–6489

    2019

  5. [14]

    Agentformer: Agent-aware transformers for socio-temporal multi-agent forecasting,

    Y. Yuan, X. Weng, Y. Ou, and K. Kitani, “Agentformer: Agent-aware transformers for socio-temporal multi-agent forecasting,” in Proceedings of the IEEE/CVF International Conference on Computer Vi- sion (ICCV), 2021

    2021

  6. [15]

    You’ll never walk alone: Modeling social behavior for multi-target tracking,

    S. Pellegrini, A. Ess, K. Schindler, and L. van Gool, “You’ll never walk alone: Modeling social behavior for multi-target tracking,” in 2009 IEEE 12th International Conference on Computer Vision, 2009, pp. 261–268

    2009

  7. [16]

    Crowds by example,

    A. Lerner, Y. Chrysanthou, and D. Lischinski, “Crowds by example,” Computer Graphics Forum, vol. 26, no. 3, pp. 655–664, 2007. [Online]. Available: https://onlinelibrary.wiley.com/doi/abs/10.1111/ j.1467-8659.2007.01089.x

    2007

  8. [17]

    Pie: A large-scale dataset and models for pedes- trian intention estimation and trajectory prediction,

    A. Rasouli, I. Kotseruba, T. Kunic, and J. Tsotsos, “Pie: A large-scale dataset and models for pedes- trian intention estimation and trajectory prediction,” in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 6261–6270. 12

    2019

  9. [18]

    Leapfrog diffusion model for stochastic trajectory prediction,

    W. Mao, C. Xu, Q. Zhu, S. Chen, and Y. Wang, “Leapfrog diffusion model for stochastic trajectory prediction,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2023, pp. 5517–5526

    2023

  10. [19]

    Remember intentions: Retrospective-memory-based trajec- tory prediction,

    C. Xu, W. Mao, W. Zhang, and S. Chen, “Remember intentions: Retrospective-memory-based trajec- tory prediction,” pp. 6488–6497, June 2022

    2022

  11. [20]

    Social force model for pedestrian dynamics,

    D. Helbing and P. Molnar, “Social force model for pedestrian dynamics,” Physical review E, vol. 51, no. 5, p. 4282, 1995

    1995

  12. [21]

    You’ll never walk alone: Modeling social behavior for multi-target tracking,

    S. Pellegrini, A. Ess, K. Schindler, and L. Van Gool, “You’ll never walk alone: Modeling social behavior for multi-target tracking,” in 2009 IEEE 12th international conference on computer vision. IEEE, 2009, pp. 261–268

    2009

  13. [22]

    Social lstm: Human trajectory prediction in crowded spaces,

    A. Alahi, K. Goel, V. Ramanathan, A. Robicquet, L. Fei-Fei, and S. Savarese, “Social lstm: Human trajectory prediction in crowded spaces,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2016

    2016

  14. [23]

    Sti-gan: Multimodal pedestrian trajectory predic- tion using spatiotemporal interactions and a generative adversarial network,

    L. Huang, J. Zhuang, X. Cheng, R. Xu, and H. Ma, “Sti-gan: Multimodal pedestrian trajectory predic- tion using spatiotemporal interactions and a generative adversarial network,” IEEE Access, vol. 9, pp. 50 846–50 856, 2021

    2021

  15. [24]

    Haar wavelet based block autoregressive flows for trajectories,

    A. Bhattacharyya, C. Straehle, M. Fritz, and B. Schiele, “Haar wavelet based block autoregressive flows for trajectories,” CoRR, vol. abs/2009.09878, 2020. [Online]. Available: https://arxiv.org/abs/2009.09878

    2009 arXiv

  16. [25]

    Trajectron++: Dynamically-feasible trajec- tory forecasting with heterogeneous data,

    T. Salzmann, B. Ivanovic, P. Chakravarty, and M. Pavone, “Trajectron++: Dynamically-feasible trajec- tory forecasting with heterogeneous data,” inComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XVIII 16. Springer, 2020, p...

    2020

  17. [26]

    Gohome: Graph-oriented heatmap output for future motion estimation,

    T. Gilles, S. Sabatini, D. Tsishkou, B. Stanciulescu, and F. Moutarde, “Gohome: Graph-oriented heatmap output for future motion estimation,” arXiv preprint arXiv:2109.01827, 2021

    2021 arXiv

  18. [27]

    Social- bigat: Multimodal trajectory forecasting using bicycle-gan and graph attention networks,

    V. Kosaraju, A. Sadeghian, R. Mart ´ ın-Mart ´ ın, I. D. Reid, S. H. Rezatofighi, and S. Savarese, “Social- bigat: Multimodal trajectory forecasting using bicycle-gan and graph attention networks,” in NeurIPS, 2019

    2019

  19. [28]

    Dada: Driver attention prediction in driving accident scenarios,

    J. Fang, D. Yan, J. Qiao, J. Xue, and H. Yu, “Dada: Driver attention prediction in driving accident scenarios,” IEEE Transactions on Intelligent Transportation Systems, 2021

    2021

  20. [29]

    Loki: Long term and key intentions for trajectory prediction,

    H. Girase, H. Gang, S. Malla, J. Li, A. Kanehara, K. Mangalam, and C. Choi, “Loki: Long term and key intentions for trajectory prediction,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 9803–9812

    2021

  21. [30]

    Ptp: Parallelized tracking and prediction with graph neural networks and diversity sampling,

    X. Weng, Y. Yuan, and K. Kitani, “Ptp: Parallelized tracking and prediction with graph neural networks and diversity sampling,” IEEE Robotics and Automation Letters, vol. 6, pp. 4640–4647, 7 2021

    2021

  22. [31]

    Large-scale graph representation learning with very deep gnns and self-supervision,

    R. Addanki, P. W. Battaglia, D. Budden, A. Deac, J. Godwin, T. Keck, W. L. S. Li, A. Sanchez- Gonzalez, J. Stott, S. Thakoor et al., “Large-scale graph representation learning with very deep gnns and self-supervision,” arXiv preprint arXiv:2107.09422, 2021

    2021 arXiv

  23. [32]

    Spatiotemporal relation- ship reasoning for pedestrian intent prediction,

    B. Liu, E. Adeli, Z. Cao, K.-H. Lee, A. Shenoi, A. Gaidon, and J. C. Niebles, “Spatiotemporal relation- ship reasoning for pedestrian intent prediction,” IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 3485–3492, 2020

    2020

  24. [33]

    Benchmark for evaluating pedestrian action prediction,

    I. Kotseruba, A. Rasouli, and J. K. Tsotsos, “Benchmark for evaluating pedestrian action prediction,” in 2021 IEEE Winter Conference on Applications of Computer Vision (WACV), 2021, pp. 1257–1267

    2021

  25. [34]

    A comprehensive intention prediction method considering vehicle interaction,

    W. Cai, G. He, J. Hu, H. Zhao, Y. Wang, and B. Gao, “A comprehensive intention prediction method considering vehicle interaction,” 2020 4th CAA International Conference on Vehicular Control and Intelligence, CVCI 2020, pp. 204–209, 12 2020

    2020

  26. [35]

    Transformer networks for trajectory forecasting,

    F. Giuliari, I. Hasan, M. Cristani, and F. Galasso, “Transformer networks for trajectory forecasting,” in 2020 25th international conference on pattern recognition (ICPR). IEEE, 2021, pp. 10 335–10 342

    2020

  27. [36]

    S2tnet: Spatio-temporal transformer networks for trajectory prediction in autonomous driving,

    W. Chen, F. Wang, and H. Sun, “S2tnet: Spatio-temporal transformer networks for trajectory prediction in autonomous driving,” in Asian Conference on Machine Learning. PMLR, 2021, pp. 454–469. 13

    2021

  28. [37]

    Euro-pvi: Pedestrian vehicle interactions in dense urban centers,

    A. Bhattacharyya, D. O. Reino, M. Fritz, and B. Schiele, “Euro-pvi: Pedestrian vehicle interactions in dense urban centers,” 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 6404–6413, 2021

    2021

  29. [38]

    Social influence analysis in large-scale networks,

    J. Tang, J. Sun, C. Wang, and Z. Yang, “Social influence analysis in large-scale networks,” in Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ser. KDD ’09. New York, NY, USA: Association for Computing Machinery, 2009, p. 807...

    2009

  30. [39]

    Multi-scale attributed node embedding,

    B. Rozemberczki, C. Allen, and R. Sarkar, “Multi-scale attributed node embedding,”Journal of Complex Networks, vol. 9, no. 2, p. cnab014, 2021

    2021

  31. [40]

    Attending to graph transformers,

    L. M¨ uller, M. Galkin, C. Morris, and L. Ramp´ aˇ sek, “Attending to graph transformers,”arXiv preprint arXiv:2302.04181, 2023

    2023 arXiv

  32. [41]

    U-net: Convolutional networks for biomedical image seg- mentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image seg- mentation,” in Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18....

    2015

  33. [42]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. u. Kaiser, and I. Polosukhin, “Attention is all you need,” in Advances in Neural Information Processing Systems, I. Guyon, U. V. Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnet...

    2017

  34. [43]

    Graph neural networks with learnable structural and positional representations,

    V. P. Dwivedi, A. T. Luu, T. Laurent, Y. Bengio, and X. Bresson, “Graph neural networks with learnable structural and positional representations,” arXiv preprint arXiv:2110.07875, 2021

    2021 arXiv

  35. [44]

    Social attention: Modeling attention in human crowds,

    A. Vemula, K. Muelling, and J. Oh, “Social attention: Modeling attention in human crowds,” in 2018 IEEE international Conference on Robotics and Automation (ICRA). IEEE, 2018, pp. 4601–4607

    2018

  36. [45]

    Social gan: Socially acceptable trajectories with generative adversarial networks,

    A. Gupta, J. Johnson, L. Fei-Fei, S. Savarese, and A. Alahi, “Social gan: Socially acceptable trajectories with generative adversarial networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 2255–2264

    2018

  37. [46]

    Stepwise goal-driven networks for trajectory predic- tion,

    C. Wang, Y. Wang, M. Xu, and D. J. Crandall, “Stepwise goal-driven networks for trajectory predic- tion,” IEEE Robotics and Automation Letters, vol. 7, no. 2, pp. 2716–2723, 2022

    2022

  38. [47]

    Collaborative motion prediction via neural motion message passing,

    Y. Hu, S. Chen, Y. Zhang, and X. Gu, “Collaborative motion prediction via neural motion message passing,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 6319–6328

    2020

  39. [49]

    It is not the journey but the destination: Endpoint conditioned trajectory prediction,

    K. Mangalam, H. Girase, S. Agarwal, K.-H. Lee, E. Adeli, J. Malik, and A. Gaidon, “It is not the journey but the destination: Endpoint conditioned trajectory prediction,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part...

    2020

  40. [50]

    Groupnet: Multiscale hypergraph neural networks for trajectory prediction with relational reasoning,

    C. Xu, M. Li, Z. Ni, Y. Zhang, and S. Chen, “Groupnet: Multiscale hypergraph neural networks for trajectory prediction with relational reasoning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2022, pp. 6498–6507

    2022

  41. [51]

    Stochastic trajectory prediction via motion indeterminacy diffusion,

    T. Gu, G. Chen, J. Li, C. Lin, Y. Rao, J. Zhou, and J. Lu, “Stochastic trajectory prediction via motion indeterminacy diffusion,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2022, pp. 17 113–17 122

    2022

  42. [52]

    Egocentric vision-based future vehicle localization for intelligent driving assistance systems,

    Y. Yao, M. Xu, C. Choi, D. J. Crandall, E. M. Atkins, and B. Dariush, “Egocentric vision-based future vehicle localization for intelligent driving assistance systems,” in 2019 International Conference on Robotics and Automation (ICRA). IEEE, 2019, pp. 9711–9717

    2019

  43. [53]

    Future person localization in first-person videos,

    T. Yagi, K. Mangalam, R. Yonetani, and Y. Sato, “Future person localization in first-person videos,” 2018. 14

    2018

  44. [54]

    Long-term on-board prediction of people in traffic scenes under uncertainty,

    A. Bhattacharyya, M. Fritz, and B. Schiele, “Long-term on-board prediction of people in traffic scenes under uncertainty,” 2018

    2018

  45. [55]

    Pedestrian behavior prediction via multitask learning and categorical interaction modeling,

    A. Rasouli, M. Rohani, and J. Luo, “Pedestrian behavior prediction via multitask learning and categorical interaction modeling,” CoRR, vol. abs/2012.03298, 2020. [Online]. Available: https://arxiv.org/abs/2012.03298

    2012 arXiv

  46. [56]

    Auto-encoding variational bayes,

    D. P. Kingma and M. Welling, “Auto-encoding variational bayes,” arXiv preprint arXiv:1312.6114, 2013

    2013 arXiv

  47. [57]

    Pedestrian trajectory prediction with convolutional neural networks,

    S. Zamboni, Z. T. Kefato, S. Girdzijauskas, C. Nor´ en, and L. Dal Col, “Pedestrian trajectory prediction with convolutional neural networks,” Pattern Recognition, vol. 121, p. 108252, 2022

    2022

  48. [58]

    Social-stgcnn: A social spatio-temporal graph convolutional neural network for human trajectory prediction,

    A. Mohamed, K. Qian, M. Elhoseiny, and C. Claudel, “Social-stgcnn: A social spatio-temporal graph convolutional neural network for human trajectory prediction,” in Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, 2020, pp. 14 424–14 432

    2020

  49. [59]

    Spatio-temporal graph transformer networks for pedestrian trajectory prediction,

    C. Yu, X. Ma, J. Ren, H. Zhao, and S. Yi, “Spatio-temporal graph transformer networks for pedestrian trajectory prediction,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XII 16. Springer, 2020, pp. 507–523

    2020

  50. [60]

    Towards interpretable semantic segmentation via gradient-weighted class activation mapping (student abstract),

    K. Vinogradova, A. Dibrov, and G. Myers, “Towards interpretable semantic segmentation via gradient-weighted class activation mapping (student abstract),” Proceedings of the AAAI, vol. 34, no. 10, p. 13943–13944, Apr. 2020. [Online]. Available: http://dx.doi.org/10.1609/aaai.v3...

    2020 doi

  51. [61]

    Grad-cam: Visual expla- nations from deep networks via gradient-based localization,

    R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-cam: Visual expla- nations from deep networks via gradient-based localization,” in Proceedings of the IEEE international conference on computer vision, 2017, pp. 618–626. 15 A Conditional Variati...

    2017

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.