CoChat
Sign Up
Cited by 60Space Satellite Systems and ControlAdvanced Neural Network ApplicationsAdvanced Image and Video Retrieval Techniques

Revisiting Monocular Satellite Pose Estimation With Transformer

Wang, Zi, Zhang, Zhuo, Sun, Xiaoliang, Li, Zhang, Yu, Qifeng·2022·IEEE Transactions on Aerospace and Electronic Systems
DOI 10.1109/taes.2022.3161605
Open this paper in CoChat — free account

Abstract

Convolutional neural networks (CNNs) have been adopted in monocular satellite pose estimation and achieve superior performance over traditional methods. However, existing CNN-based methods suffer from bias toward texture, indirect description of absolute distance, and lack of long-range dependence modeling. Such factors limit the generalizability of CNN-based methods. Motivated by the striking achievements of transformer models, this article adopts transformer blocks for satellite pose estimation from a single RGB image, proposing an efficient monocular satellite pose estimation method. First, we design an effective satellite representation model based on a set of keypoints. Then, considering monocular satellite pose estimation characteristics, we construct an end-to-end keypoint-set prediction network and build the bipartite loss function. Further, we improve the backbone structure for high-quality feature extraction. Experimental results on a public benchmark dataset indicate that the proposed method achieves second and third place on the synthetic and real test sets, respectively, using only synthetic training data. We also demonstrate that our keypoint predictor takes half as much time as the first-placed method in our comparison, and therefore achieves a better tradeoff between speed and accuracy than existing approaches.

Integrity signals

Registry

Crossref · 10.1109/taes.2022.3161605

Cites

43 of 43 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Wang et al.'s transformer-based method consumes a lot of hardware resources and has more than 200 million trainable parameters, making it hard to use for space applications.

    “This work consumes a lot of hardware resources and has more than 200 million trainable parameters that make this architecture hard to use for space applications.”

    From Transformer Network-Aided Relative Pose Estimation for Non-cooperative Spacecraft Using Vision Sensor (2024) · cites this paper as [17]

  • Wang et al. [17] proposed a transformer architecture based on the DETR transformer.

    “Wang et al. [17] proposed a transformer architecture based on the DETR transformer.”

    From Transformer Network-Aided Relative Pose Estimation for Non-cooperative Spacecraft Using Vision Sensor (2024) · cites this paper as Wang et al. [17]

  • A PnP-based approach reported 150 ms per image (~7 FPS) on an NVIDIA RTX 2080Ti GPU and an i7-7700K CPU.

    “For example, one PnP-based approach [35] reported 150 ms per image ( ∼ 7 FPS) on a high-performance NVIDIA RTX 2080Ti GPU (peak draw 250 Watts) and an i7-7700K CPU, far exceeding the typical power budgets of a spacecraft ( ∼ 1-25 Watts).”

    From FastPose-ViT: A Vision Transformer for Real-Time Spacecraft Pose Estimation (2026) · cites this paper as [35]

  • Existing state-of-the-art spacecraft pose estimation methods often rely on iterative PnP-based algorithms, which are computationally intensive and ill-suited for real-time deployment on resource-constrained edge devices.

    “Existing state-of-the-art methods often rely on iterative Perspective-n-Point (PnP)-based algorithms, which are computationally intensive and ill-suited for real-time deployment on resource-constrained edge devices.”

    From FastPose-ViT: A Vision Transformer for Real-Time Spacecraft Pose Estimation (2026) · cites this paper as [35]

  • Previous satellite pose estimation approaches represent a satellite as a set of 2D keypoints and then estimate the satellite pose using PnP algorithms.

    “Meanwhile, previous satellite pose estimation approaches [12], [15] represent a satellite as a set of 2D keypoints, and then estimate the satellite pose using the perspective-n-point (PnP) algorithms [27].”

    From Bridging the Domain Gap in Satellite Pose Estimation: A Self-Training Approach Based on Geometrical Constraints (2024) · cites this paper as [15]

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Wang, Z., Zhang, Z., Sun, X., Li, Z., & Yu, Q.
Affiliations
National University of Defense Technology (CN)
ORCID
Venue
IEEE Transactions on Aerospace and Electronic Systems
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Published
2022-10
Volume
58
Issue
5
Pages
4279-4294
ISSN
  • 0018-9251 (ISSN-L)
  • 1557-9603
  • 2371-9877
DOI
10.1109/taes.2022.3161605
OpenAlex
W4226098472
Cited by
59
References
43
Citations by year
2026: 14 · 2025: 23 · 2024: 11 · 2023: 10 · 2022: 1
FWCI
5.01
Type
journal article
Language
en
Open access
closed
Licence
Locations
IEEE Transactions on Aerospace and Electronic Systems (publishedVersion)
Funders
  • National Natural Science Foundation of China (62003357, 61801491)
  • Natural Science Foundation of Hunan Province (2019JJ50732, 2019JJ50728)
  • Postgraduate Scientific Research Innovation Project of Hunan Province (CX20200024, CX20200025, CX20200088)
Topics
  • Physical Sciences › Engineering › Aerospace Engineering › Space Satellite Systems and Control
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
Keywords
transformer models, synthetic training data, speed-accuracy trade-off

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review