Pith. sign in

REVIEW 4 major objections 6 minor 39 references

Motion Aware ViT-based Framework for Monocular 6-DoF Spacecraft Pose Estimation

T0 review · 4 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Using motion between video frames improves monocular 6-DoF spacecraft pose estimation over single-image baselines.

desk verdict A clean domain transfer of MTPose to spacecraft pose estimation with decent results, but the 2D claim is undercut by a non-standard PCK normalization and the eval lacks baselines. read the letter →

arxiv 2509.06000 v1 pith:2SXUOC6O submitted 2025-09-07 cs.CV

Jose Sosa , Dan Pineau , Arunkumar Rathinam , Abdelrahman Shabayek , Djamila Aouada This is my paper · ORCID
classification cs.CV
keywords 6-DoFposeestimationspacecraftmotion-awareheatmapsopticalflowVisionTransformerkeypointlocalizationtemporalvideoinformationSPADES-RGB
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 argues that monocular spacecraft pose estimation should use the motion visible across video frames rather than treating each image independently. It adapts a motion-aware keypoint regression technique from human pose estimation, representing each of eight spacecraft keypoints as an elliptical heatmap whose orientation and stretch encode how that keypoint moves between the central frame and its two neighbours. Optical flow supplies the motion, a Vision Transformer encodes the central frame, and cross-attention and deformable convolutions fuse the two into a final heatmap; a Perspective-n-Point solver then recovers the 6-DoF pose. On SPADES-RGB the motion-aware model reaches 94.05% PCK@10 and pose error 0.157, versus 92.63% and 0.197 for the same model without motion, and it transfers without fine-tuning to the synthetic and real subsets of SPARK-2024.

What carries the argument

The motion-aware heatmap is the load-bearing object. It turns the temporal signal into a per-keypoint training target: instead of a fixed circular Gaussian centred on the keypoint, each heatmap is a 2D elliptical Gaussian whose rotation angle and axis lengths are derived from the keypoint's displacement between the central frame and an adjacent frame; static keypoints fall back to circular Gaussians. Because the two pairwise heatmaps and the final fused heatmap are all supervised, the network is forced to learn where the keypoint is and how it is moving, and the PnP stage then converts that localisation into a 6-DoF pose.

What would settle it

Train the motion-aware model and the single-frame baseline on an augmented version of SPADES-RGB where the dominant motion is in-plane rotation about the camera's optical axis: the motion-aware model should improve, since the 2D ellipse encoding is well-matched. Then construct sequences dominated by out-of-plane tumbling and check whether the motion-aware model's PCK@10 falls below the single-frame baseline; that would indicate the 2D Gaussian encoding misrepresents the true 3D motion.

Watch

Extended reading notes

Core claim

The paper's central claim is that explicit motion cues, injected at the heatmap level, are what make temporal information useful for keypoint-based spacecraft pose estimation. On SPADES-RGB the full model reaches PCK@10 of 94.05% and pose error 0.157, against 92.63% and 0.197 for the same model without motion. The mechanism is the motion-aware heatmap: each keypoint is an elliptical Gaussian whose orientation and elongation encode that keypoint's displacement between the central and adjacent frames; optical flow supplies the displacement, cross-attention fuses it with ViT features, and deformable convolutions merge the pairwise heatmaps into the final prediction. Freezing the ViT encoder des

Load-bearing premise

The motion-aware heatmap borrows its parameterization from human pose estimation, assuming a keypoint's apparent motion between two frames can be encoded as a 2D elliptical Gaussian with a displacement and a rotation angle; if a spacecraft rotates mostly out of the image plane, that 2D encoding may not align with the true perspective-projected motion, and the added cues could hurt rather than help.

Editorial extensions

If this is right

  • On the SPADES-RGB benchmark, adding motion through three-frame input improves 2D keypoint localisation (PCK@10 94.05 vs 92.63) and the composite 6-DoF pose error (E_P 0.157 vs 0.197) over the single-frame baseline.
  • Because the final stage is still a standard keypoint-plus-PnP pipeline, the motion-aware heatmap head can in principle be attached to other keypoint-based spacecraft pose estimators without changing their pose solvers.
  • The model generalises to SPARK-2024 synthetic data without fine-tuning (E_P 0.099, better than on SPADES-RGB), and retains non-trivial performance on real imagery (E_P 0.320), suggesting the motion encoding transfers across rendering and capture conditions.
  • Freezing the ViT encoder fails badly (E_P 2.37), indicating that domain-specific fine-tuning is required for the motion-aware scheme to work on spacecraft, unlike in the human-pose setting that inspired it.

Reading between the lines

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

  • The 2D elliptical-Gaussian motion encoding comes from human pose estimation, where motion is mostly in-plane; spacecraft can tumble with large out-of-plane rotation, so a natural extension is to parameterise the heatmap from the projected 3D rotational flow rather than the 2D displacement. The paper's seven-frame sampling interval suggests the encoding needs enough apparent motion to work.
  • The per-sequence PCK analysis shows many SPADES-RGB sequences saturate near 90% correct keypoints, so the measured gap between motion-aware and single-image models may understate the difference; evaluating on sequences with faster rotation or occlusion would separate the two more sharply.
  • Because the pipeline assumes bounding-box crops, the method is positioned for tracking use; coupling it with a detector and testing on full-frame video would show whether the motion cues survive crop noise.
  • A stronger test of the motion encoding is to replace the motion-aware elliptical heatmaps with a plain multi-frame input (for example, concatenating the three frames) in the same architecture; if the gain disappears, it is the elliptical parameterization, not the extra context, that carries the improvement.
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 / 6 minor

Summary. The manuscript proposes a motion-aware keypoint-based framework for monocular 6-DoF spacecraft pose estimation. It adapts MTPose (Song et al., IJCAI 2024) to the spacecraft domain: three temporally sampled frames are input, the central frame is encoded with a ViT, a frozen RAFT network estimates optical flow to the two adjacent frames, and multi-head cross-attention fuses the motion vectors with image features. Decoders produce pairwise motion-aware heatmaps, which are merged via deformable convolutions into the final heatmap. Keypoints are extracted by argmax, and PnP with RANSAC recovers the 6-DoF pose. Experiments on SPADES-RGB report PCK@10 of 94.05 and pose error E_P=0.157, versus an N-MA self-baseline of 92.63 and 0.197. Zero-shot evaluation on SPARK-2024 synthetic and real data shows the expected domain-shift degradation. The paper claims improved performance over single-image baselines and sets a benchmark on SPADES-RGB.

Significance. If the evaluation issues are addressed, the paper offers a useful temporal alternative for spacecraft pose estimation, a domain where few learning-based temporal methods exist. Strengths include a clear self-ablation showing that motion information improves both keypoint localization and pose accuracy, zero-shot cross-dataset evaluation, and use of a standard PnP pipeline. The novelty is mainly the transfer of MTPose to the spacecraft domain; the reported gains are modest but plausible. However, the evidence is currently overstated: the PCK normalization is non-standard, no external single-image baselines are provided, and no error bars or statistical tests support the reported differences.

major comments (4)
  1. [§4.2, Metrics] The PCK normalization is defined as the diagonal of the bounding box enclosing the predicted keypoints. Standard PCK uses the ground-truth object bounding box (or a fixed object-scale reference). With predicted-box normalization, a model that overestimates the spread of keypoints is given a larger tolerance, so the reported MA vs N-MA differences in PCK@5/10 (84.05 vs 70.33 and 94.05 vs 92.63) may not reflect genuine localization accuracy. Please re-evaluate with a ground-truth-based normalization and report both values. The abstract's 2D-localization claim rests on this metric.
  2. [§4.4 and Table 1] The paper claims a 'benchmark' on SPADES-RGB and 'improved performance over single-image baselines', but the only comparison is the self-ablation N-MA in Table 2. No existing single-image SPE method is evaluated on the same split, and no error bars or statistical tests across the 45 test sequences are given. The 6-DoF errors (E_t=0.227, E_R=7.57, E_P=0.157) are point estimates. Please add at least one or two standard single-image baselines (e.g., HRNet/ResNet keypoint or a published SPE method) trained on the same data, and report standard deviations or confidence intervals; otherwise the central claim is not established beyond the self-ablation.
  3. [§3.1 and §4.3] The method delegates key implementation details to [8]: 'Due to space constraints, we refer readers to [8] for complete implementation details' (Sec. 3.1) and 'All other implementation details ... follow the setup in [8]' (Sec. 4.3). This is not reproducible for the spacecraft domain: the keypoint definitions, the static-keypoint threshold, the seven-frame sampling interval, the loss weights between final and auxiliary heatmaps, and the exact Gaussian parameterization are not stated. Since MTPose was designed for human pose estimation, the adaptation is non-trivial. Please include the missing formulas and hyperparameters in the paper or a supplement.
  4. [§3.1, motion model assumption] The motion-aware heatmap encodes keypoint displacement as a 2D elliptical Gaussian with a rotation angle. This model is built for human body motion, where keypoint flow is largely 2D in the image. For a spacecraft in arbitrary 6-DoF motion, apparent keypoint motion is the perspective projection of 3D rotation; out-of-plane rotation can produce keypoint movement that the 2D motion model may not represent. The paper does not include an ablation that isolates the motion-aware heatmap parameterization from the presence of flow information (e.g., a variant that simply concatenates RAFT features into the decoder). Such an experiment would confirm that the specific motion encoding, not just additional input channels, drives the improvement.
minor comments (6)
  1. [§4.5, Table 2] The text refers to the FE variant as 'the second row' twice; it is actually the third row of Table 2. Please correct.
  2. [§4.2 and Tables 1-2] Metric names are inconsistent: Eq. (2) defines E_q, but Tables 1-2 use E_R; E_t is also written inconsistently. Unify the notation.
  3. [§4.4] Typo: 'Futhermore' should be 'Furthermore'.
  4. [§4.3 and References] The text says 'the original model for human pose estimation by Song et al. [39]' and cites [39], which is a spacecraft navigation survey by Jianing Song et al., not the IJCAI MTPose paper [8]. The citation should be to [8] (Inpyo Song et al.).
  5. [Table 3] Filtering by PCK@10 and then reporting PCK@10 on the retained subset is tautological; the 100.00 value for the PCK>90 row is especially surprising and should be explained. The 81% statistic is informative, but the follow-on pose metrics on filtered subsets are not a valid ablation.
  6. [§4.4] PCK@1 results (18.17 on SPADES) are reported but not discussed. A brief interpretation would help readers calibrate the strict-threshold performance.

Circularity Check

1 steps flagged · score 4.0 of 10

One self-referential PCK metric partially undermines the 2D-localisation comparison; the 6-DoF metrics and method are otherwise independent.

  1. self definitional [Section 4.2, Metrics (PCK definition)]
    "Following standard practices on 2D keypoint localisation, the normalisation factor is defined as the diagonal length of the bounding box enclosing the predicted keypoints."

    PCK thresholds are normalized by the diagonal of the bounding box enclosing the predicted keypoints. Thus the tolerance that decides whether a predicted keypoint is correct is computed from the model's own predictions: larger predicted spread yields a larger acceptance radius for the same absolute error. The reported MA vs N-MA PCK@10 difference (94.05 vs 92.63) is therefore not a fixed-accuracy comparison; part of the score is determined by the predictor's output distribution. The 2D-localisation claim is evaluated with a criterion defined in terms of the very quantity being judged.

full rationale

The motion-aware framework itself is not circular: the model is trained on held-out SPADES-RGB sequences with ground-truth heatmaps and optical flow from a frozen external RAFT model; the reported pose errors (Et, Eq, EP) use standard formulas independent of the model's outputs; the ablation baseline N-MA is the same architecture without temporal inputs; and the SPARK-2024 results are zero-shot on a public benchmark. No parameter is fitted to the test set. The only genuinely self-referential element is the PCK normalisation in Sec. 4.2: the threshold is defined as a fraction of the diagonal of the predicted keypoints' bounding box. This makes the 2D-localisation comparison partially dependent on the predictor's own spread, so MA vs N-MA PCK differences cannot be interpreted as a pure accuracy gain. This is a metric-construction issue rather than a fitted-input or self-citation chain, and the 6-DoF evidence remains independent. The reliance on the authors' own SPADES/SPARK datasets is a provenance overlap, not evidence that the results are baked into the method; the datasets are public and the paper reports zero-shot transfer. The omission of the exact Gaussian heatmap formulation with a pointer to [8] is an external implementation detail, not a circular step.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the transferability of MTPose's motion-aware heatmap machinery to spacecraft imagery, plus standard assumptions about known 3D models and bounding boxes. No new physical entities are introduced.

free parameters (3)
  • Seven-frame sampling interval = 7
    Chosen by hand (Section 4.3) to encourage meaningful motion; not justified across datasets.
  • Static keypoint threshold for motion encoding = not specified (from [8])
    Determines when to fall back to circular Gaussians; value inherited from human pose model, not re-tuned for spacecraft.
  • Loss weighting between final and auxiliary heatmaps = not specified (from [8])
    Deferred to [8]; could affect training balance.
assumptions (4)
  • domain assumption Known 2D-3D keypoint correspondences from the spacecraft CAD model are correct
    PnP stage (Section 3.2) requires exact 3D keypoint locations; any modeling error propagates directly to pose.
  • domain assumption Bounding box annotations are available at test time
    Section 4.3 states the cropped region is centred on the bounding box, which is assumed provided; real systems would need a detector.
  • domain assumption Pre-trained RAFT optical flow transfers to spacecraft imagery
    Eq. (1) uses RAFT frozen; no evidence that flow estimates on synthetic spacecraft images are accurate.
  • domain assumption Motion-aware heatmap parameterization from human pose estimation is appropriate for spacecraft keypoints
    Section 3.1: '[w]e follow the same motion-aware heatmap generation strategy proposed by [8]' without domain adaptation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Motion Aware ViT-based Framework for Monocular 6-DoF Spacecraft Pose Estimation." pith.science (2026). https://pith.science/paper/2SXUOC6O

@misc{pith2026250906000,
  author       = {Pith},
  title        = {Pith review of: Motion Aware ViT-based Framework for Monocular 6-DoF Spacecraft Pose Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2SXUOC6O}},
  note         = {Machine review of arXiv:2509.06000}
}
read the original abstract

Monocular 6-DoF pose estimation plays an important role in multiple spacecraft missions. Most existing pose estimation approaches rely on single images with static keypoint localisation, failing to exploit valuable temporal information inherent to space operations. In this work, we adapt a deep learning framework from human pose estimation to the spacecraft pose estimation domain that integrates motion-aware heatmaps and optical flow to capture motion dynamics. Our approach combines image features from a Vision Transformer (ViT) encoder with motion cues from a pre-trained optical flow model to localise 2D keypoints. Using the estimates, a Perspective-n-Point (PnP) solver recovers 6-DoF poses from known 2D-3D correspondences. We train and evaluate our method on the SPADES-RGB dataset and further assess its generalisation on real and synthetic data from the SPARK-2024 dataset. Overall, our approach demonstrates improved performance over single-image baselines in both 2D keypoint localisation and 6-DoF pose estimation. Furthermore, it shows promising generalisation capabilities when testing on different data distributions.

Figures

Figures reproduced from arXiv: 2509.06000 by the authors.

Figure 1
Figure 1. Motion-aware framework for monocular 6- DoF SPE. The model first predicts 2D keypoint locations relying on motion cues from image sequences. Then it uses a PnP solver to recover the full 6-DoF pose. sequential data. Recent research has shown that incorpo￾rating temporal information can improve pose accuracy and enhance robustness [3, 4, 5, 6, 7]. However, the low diversity of temporal training data in the space doma… view at source ↗
Figure 2
Figure 2. Overview of the proposed method for 2D keypoint localisation. We input three temporally adjacent images [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Qualitative results for 2D keypoint localisa [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Distribution of PCK@10 scores per each se [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 37 canonical work pages

  1. [8]

    Motion-aware heatmap regres- sion for human pose estimation in videos

    Inpyo Song et al. “Motion-aware heatmap regres- sion for human pose estimation in videos”. In: Proceedings of the Thirty-third International Joint Conference on Artificial Intelligence, IJCAI 2024. IJCAI. 2024, pp. 1245–1253

    work page 2024

  2. [39]

    Deep learning-based spacecraft relative naviga- tion methods: A survey

    Jianing Song, Duarte Rondao, and Nabil Aouf. “Deep learning-based spacecraft relative naviga- tion methods: A survey”. In:Acta Astronautica 191 (2022), pp. 22–40

    work page 2022

  3. [1]

    A review of cooper- ative and uncooperative spacecraft pose determi- nation techniques for close-proximity operations

    Roberto Opromolla et al. “A review of cooper- ative and uncooperative spacecraft pose determi- nation techniques for close-proximity operations”. In:Progress in Aerospace Sciences93 (2017), pp. 53–72

    work page 2017

  4. [2]

    A survey on deep learning- based monocular spacecraft pose estimation: Cur- rent state, limitations and prospects

    Leo Pauly et al. “A survey on deep learning- based monocular spacecraft pose estimation: Cur- rent state, limitations and prospects”. In:Acta As- tronautica(2023)

    work page 2023

  5. [3]

    Chinet: Deep recurrent convolutional learn- ing for multimodal spacecraft pose estimation

    Duarte Rondao, Nabil Aouf, and Mark A Richard- son. “Chinet: Deep recurrent convolutional learn- ing for multimodal spacecraft pose estimation”. In: IEEE Transactions on Aerospace and Electronic Systems59.2 (2022), pp. 937–949

    work page 2022

  6. [4]

    CubeSat- CDT: A Cross-Domain Dataset for 6-DoF Tra- jectory Estimation of a Symmetric Spacecraft

    Mohamed Adel Mohamed Ali et al. “CubeSat- CDT: A Cross-Domain Dataset for 6-DoF Tra- jectory Estimation of a Symmetric Spacecraft”. In:Proceedings of the 17th European Conference on Computer Vision Workshops (ECCVW 2022). 2022

    work page 2022

  7. [5]

    Leveraging tem- poral information for 3d trajectory estimation of space objects

    Mohamed Adel Musallam et al. “Leveraging tem- poral information for 3d trajectory estimation of space objects”. In:Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021, pp. 3816–3822

    work page 2021

  8. [6]

    CroSpace6D: Leveraging Geometric and Motion Cues for High-Precision Cross-Domain 6DoF Pose Estimation for Non- Cooperative Spacecrafts

    Jianhong Zuo et al. “CroSpace6D: Leveraging Geometric and Motion Cues for High-Precision Cross-Domain 6DoF Pose Estimation for Non- Cooperative Spacecrafts”. In:Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024, pp. 6857–6863

    work page 2024

Show all 39 references
  1. [7]

    Monocular 6-DoF Pose Es- timation of Spacecrafts Utilizing Self-iterative Op- timization and Motion Consistency

    Yunfeng Zhang et al. “Monocular 6-DoF Pose Es- timation of Spacecrafts Utilizing Self-iterative Op- timization and Motion Consistency”. In:Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024, pp. 6847– 6856

    2024

  2. [9]

    Raft: Recurrent all- pairs field transforms for optical flow

    Zachary Teed and Jia Deng. “Raft: Recurrent all- pairs field transforms for optical flow”. In:Eu- ropean conference on computer vision. Springer. 2020, pp. 402–419

    2020

  3. [10]

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

    Alexey Dosovitskiy. “An image is worth 16x16 words: Transformers for image recognition at scale”. In:arXiv preprint arXiv:2010.11929 (2020)

    2010 arXiv

  4. [11]

    Spades: A realistic spacecraft pose estimation dataset using event sensing

    Arunkumar Rathinam, Haytam Qadadri, and Djamila Aouada. “Spades: A realistic spacecraft pose estimation dataset using event sensing”. In: 2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE. 2024, pp. 11760– 11766

    2024

  5. [12]

    Zenodo, Apr

    Arunkumar Rathinam et al.SPARK 2024: Datasets for Spacecraft Semantic Segmentation and Space- craft Trajectory Estimation. Zenodo, Apr. 2024. DOI:10.5281/zenodo.10908215

    2024 doi

  6. [13]

    6D object position esti- mation from 2D images: A literature review

    Giorgia Marullo et al. “6D object position esti- mation from 2D images: A literature review”. In: Multimedia Tools and Applications82.16 (2023), pp. 24605–24643

    2023

  7. [14]

    Vision-based spacecraft pose estimation via a deep convolu- tional neural network for noncooperative docking operations

    Thaweerath Phisannupawong et al. “Vision-based spacecraft pose estimation via a deep convolu- tional neural network for noncooperative docking operations”. In:Aerospace7.9 (2020), p. 126

    2020

  8. [15]

    Non-Model-Based Monoc- ular Pose Estimation Network for Uncoopera- tive Spacecraft Using Convolutional Neural Net- work

    Haoran Huang et al. “Non-Model-Based Monoc- ular Pose Estimation Network for Uncoopera- tive Spacecraft Using Convolutional Neural Net- work”. In:IEEE Sensors Journal21.21 (2021), pp. 24579–24590

    2021

  9. [16]

    Deep learning for spacecraft pose estimation from photorealistic rendering

    Pedro F Proenc ¸a and Yang Gao. “Deep learning for spacecraft pose estimation from photorealistic rendering”. In:2020 IEEE International Confer- ence on Robotics and Automation (ICRA). IEEE. 2020, pp. 6007–6013

    2020

  10. [17]

    Real-time, flight-ready, non-cooperative spacecraft pose estimation us- ing monocular imagery

    Kevin Black et al. “Real-time, flight-ready, non-cooperative spacecraft pose estimation us- ing monocular imagery”. In:arXiv preprint arXiv:2101.09553(2021)

    2021 arXiv

  11. [18]

    Wide-depth-range 6d object pose estimation in space

    Yinlin Hu et al. “Wide-depth-range 6d object pose estimation in space”. In:Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021, pp. 15870–15879

    2021

  12. [19]

    Learning-based pose estimation of non- cooperative spacecrafts with uncertainty pre- diction

    Kecen Li, Haopeng Zhang, and Chenyu Hu. “Learning-based pose estimation of non- cooperative spacecrafts with uncertainty pre- diction”. In:Aerospace9.10 (2022), p. 592

    2022

  13. [20]

    Revisiting monocular satellite pose estimation with transformer

    Zi Wang et al. “Revisiting monocular satellite pose estimation with transformer”. In:IEEE Transac- tions on Aerospace and Electronic Systems58.5 (2022), pp. 4279–4294

    2022

  14. [21]

    Investigating vision transformers for bridg- ing domain gap in satellite pose estimation

    Alessandro Lotti, Dario Modenini, and Paolo Tor- tora. “Investigating vision transformers for bridg- ing domain gap in satellite pose estimation”. In: International Conference on Applied Intelligence and Informatics. Springer. 2022, pp. 299–314

    2022

  15. [22]

    Deep residual learning for im- age recognition

    Kaiming He et al. “Deep residual learning for im- age recognition”. In:Proceedings of the IEEE con- ference on computer vision and pattern recogni- tion. 2016, pp. 770–778

    2016

  16. [23]

    Deep high-resolution representation learning for human pose estimation

    Ke Sun et al. “Deep high-resolution representation learning for human pose estimation”. In:Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019, pp. 5693– 5703

    2019

  17. [24]

    YOLO9000: better, faster, stronger

    Joseph Redmon and Ali Farhadi. “YOLO9000: better, faster, stronger”. In:Proceedings of the IEEE conference on computer vision and pattern recognition. 2017, pp. 7263–7271

    2017

  18. [25]

    Mobilenetv2: Inverted residu- als and linear bottlenecks

    Mark Sandler et al. “Mobilenetv2: Inverted residu- als and linear bottlenecks”. In:Proceedings of the IEEE conference on computer vision and pattern recognition. 2018, pp. 4510–4520

    2018

  19. [26]

    Attention is all you need

    A Vaswani. “Attention is all you need”. In:Ad- vances in Neural Information Processing Systems (2017)

    2017

  20. [27]

    Swin transformer: Hierarchical vi- sion transformer using shifted windows

    Ze Liu et al. “Swin transformer: Hierarchical vi- sion transformer using shifted windows”. In:Pro- ceedings of the IEEE/CVF international confer- ence on computer vision. 2021, pp. 10012–10022

    2021

  21. [28]

    Adaptive Neural-Network-Based Unscented Kalman Fil- ter for Robust Pose Tracking of Noncooperative Spacecraft

    Tae Ha Park and Simone D’Amico. “Adaptive Neural-Network-Based Unscented Kalman Fil- ter for Robust Pose Tracking of Noncooperative Spacecraft”. In:Journal of Guidance, Control, and Dynamics46.9 (2023), pp. 1671–1688.ISSN: 0731-5090.DOI:10.2514/1.G007387

    2023 doi

  22. [29]

    Leverag- ing Neural Network Uncertainty in Adaptive Un- scented Kalman Filter for Spacecraft Pose Esti- mation

    Lorenzo Pasqualetto Cassinis et al. “Leverag- ing Neural Network Uncertainty in Adaptive Un- scented Kalman Filter for Spacecraft Pose Esti- mation”. In:Advances in Space Research71.12 (June 15, 2023), pp. 5061–5082.ISSN: 0273-1177. DOI:10.1016/j.asr.2023.02.021

    2023 doi

  23. [30]

    Relative Pose Determination for Noncooperative Spacecraft under Non-inertial Observation Frame

    Hang Chen et al. “Relative Pose Determination for Noncooperative Spacecraft under Non-inertial Observation Frame”. In:IEEE Transactions on Aerospace and Electronic Systems(2025), pp. 1– 15.ISSN: 1557-9603.DOI:10 . 1109 / TAES . 2025.3562539

    2025

  24. [31]

    A CNN-based Archi- tecture for Relative State and Target Motion Pa- rameters Estimation in Active Debris Removal Missions

    Giuseppe Napolano et al. “A CNN-based Archi- tecture for Relative State and Target Motion Pa- rameters Estimation in Active Debris Removal Missions”. In:Acta Astronautica235 (Oct. 1, 2025), pp. 485–511.ISSN: 0094-5765.DOI:10. 1016/j.actaastro.2025.06.012

    2025

  25. [32]

    A simple neural at- tentive meta-learner

    Nikhil Mishra et al. “A simple neural at- tentive meta-learner”. In:arXiv preprint arXiv:1707.03141(2017)

    2017 arXiv

  26. [33]

    Deformable convnets v2: More deformable, better results

    Xizhou Zhu et al. “Deformable convnets v2: More deformable, better results”. In:Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019, pp. 9308–9316

    2019

  27. [34]

    A review of solutions for perspective-n-point problem in camera pose es- timation

    Xiao Xin Lu. “A review of solutions for perspective-n-point problem in camera pose es- timation”. In:Journal of Physics: Conference Series. V ol. 1087. 5. IOP Publishing. 2018, p. 052009

    2018

  28. [35]

    Proba-2 mission and new technologies overview

    Kristof Gantois et al. “Proba-2 mission and new technologies overview”. In: (2006)

    2006

  29. [36]

    Zero-G Lab: A Multi-Purpose Facility for Emulating Space Op- erations

    Miguel Olivares-Mendez et al. “Zero-G Lab: A Multi-Purpose Facility for Emulating Space Op- erations”. In:Journal of Space Safety Engineer- ing10.4 (Dec. 1, 2023), pp. 509–521.ISSN: 2468- 8967.DOI:10 . 1016 / j . jsse . 2023 . 09 . 003

    2023

  30. [37]

    SPEED+: Next-generation dataset for spacecraft pose estimation across do- main gap

    Tae Ha Park et al. “SPEED+: Next-generation dataset for spacecraft pose estimation across do- main gap”. In:2022 IEEE aerospace conference (AERO). IEEE. 2022, pp. 1–15

    2022

  31. [38]

    Imagenet: A large-scale hierar- chical image database

    Jia Deng et al. “Imagenet: A large-scale hierar- chical image database”. In:2009 IEEE conference on computer vision and pattern recognition. Ieee. 2009, pp. 248–255

    2009

Pith tools

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