Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Marginalized Generalized IoU (MGIoU): A Unified Objective Function for Optimizing Any Convex Parametric Shapes

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a single loss, MGIoU, can replace domain-specific IoU approximations for every convex parametric shape by projecting shapes onto their unique normals and averaging one-dimensional GIoU scores.

desk verdict A genuinely new, fast loss for convex shape regression with solid empirical gains, but the advertised metric properties are unproven and the IoU-correlation claim is unquantified. read the letter →

arxiv 2504.16443 v2 pith:EEPZKFND submitted 2025-04-23 cs.CV

Duy-Tho Le , Trung Pham , Jianfei Cai , Hamid Rezatofighi This is my paper · ORCID
classification cs.CV MSC 68T4568U05
keywords marginalizedgeneralizedIoUconvexshapeoptimizationorientedobjectdetection3D6-DoFestimationquadrilateraltrajectorypredictioncollisionavoidancedifferentiableloss
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's central claim is that one differentiable loss function, MGIoU, can serve as a unified objective for aligning any convex parametric shapes, replacing the patchwork of task-specific IoU approximations used across 2D rotated detection, 3D 6-DoF pose estimation, quadrilateral detection, and trajectory prediction. The proposal is to project both shapes onto their unique edge or face normals, compute a one-dimensional GIoU along each normal, and average those values; for rectangles this needs only two projections and for cuboids three. The paper argues that this simple average stays strongly correlated with true IoU while making optimization stable and cheap, and that the same projection idea, when turned toward minimizing overlap, gives a collision-avoidance loss for trajectories. If the claim holds, one loss with no task-specific balancing could replace L1/L2, Chamfer, Gaussian-based IoU approximations, and keypoint-similarity losses while cutting loss computation latency by 10-40x.

What carries the argument

The object that carries the argument is the projection set $A$ of unique directional normals obtained from the faces or edges of the two shapes, together with the one-dimensional GIoU computed on each projection. For a rectangle, edge normals come in collinear pairs, so only two directions survive; for a cuboid only three, and for an ellipse the semi-axes serve as the normals. Algorithmically, each normal produces two intervals from the min and max of the projected vertices, and the 1D GIoU of those intervals is computed in closed form; the final score is the unweighted average over $A$. The same machinery is reused three ways: maximizing the average for structured shapes (MGIoU), adding a convexity regularizer on signed distances for unstructured polygons (MGIoU+), and taking the smallest 1D GIoU across normals with a softplus to minimize overlap (MGIoU-). This one mechanism is what lets a single loss address position, size, and orientation without balancing separate terms.

What would settle it

Take a large random sample of convex shape pairs, such as rotated rectangles or cuboids, compute true volume IoU and MGIoU for each pair, and check their rank correlation. The claim would be falsified if two non-identical shapes achieve identical projections on all unique normals, giving a perfect MGIoU score, while their true IoU is well below 1, or if the overall rank correlation between MGIoU and true IoU is weak on a large sample.

Watch

Extended reading notes

Core claim

The central claim is that generalized IoU can be marginalized over a small set of shape normals without losing what matters for optimization. Concretely, for a predicted shape P and ground truth G, MGIoU takes the union of their unique directional normals, projects both shapes onto each normal, computes the one-dimensional GIoU of those two intervals, and averages. For structured shapes this average is used directly in the loss $L=(1-\mathrm{MGIoU})/2$; for unstructured shapes, a convexity regularizer is added, and for collision avoidance the smallest one-dimensional GIoU is penalized through a softplus. The paper claims this construction is symmetric, scale-invariant, zero only for identical shapes, and satisfies the triangle inequality, so the loss behaves like a metric and correlates with IoU. Empirically, the same machinery improves accuracy on oriented detection, 3D 6-DoF estimation, and quadrilateral detection, and reduces trajectory collisions by about 14 percent, while cutting loss computation latency by 10-40x compared with Gaussian and skew-IoU baselines.

Load-bearing premise

The construction depends on the assumption that a shape's few unique edge or face normals carry enough geometric information that matching all one-dimensional projections is equivalent to matching the true volumes; if two distinct convex shapes had identical projections but different true IoUs, the loss would be blind to the difference.

Editorial extensions

If this is right

  • A single loss can replace task-specific objectives such as L1/L2, Chamfer, Gaussian Wasserstein, KLD, and OKS across 2D oriented detection, 3D 6-DoF estimation, and quadrilateral detection benchmarks, with gains in the reported metrics.
  • Because only a handful of normals are needed per shape, loss computation is 10-40x faster than KFIoU/GWD/KLD and close to L1 in latency.
  • MGIoU+ permits optimizing between shapes with different vertex counts or shape families, as long as the target has more vertices, opening polygonal fitting beyond quadrilaterals.
  • MGIoU- gives a differentiable overlap-minimization term that cuts collisions in multi-agent trajectory prediction by about 14 percent with no architectural change.
  • The metric properties make the loss scale-invariant and symmetric, so it does not need rescaling or task-specific tuning when input units change.

Reading between the lines

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

  • Editorial inference: if the projection faithfulness holds beyond rectangles and cuboids, the same marginalization could give a differentiable IoU proxy for non-convex shapes by decomposing them into convex pieces; the paper does not claim this.
  • Editorial inference: a natural stress test is to compare MGIoU against true IoU on random convex polygons with many vertices; high rank correlation would confirm that the few unique normals are enough, while saturation would reveal the approximation's limit.
  • Editorial inference: the overlap-minimizing variant, which takes the smallest 1D GIoU across normals, resembles a Separating Axis Theorem certificate and could be transplanted to packing, layout, or collision-checking problems beyond trajectory prediction.
  • Editorial inference: one could replace Chamfer distance in shape registration with MGIoU+ and test whether overlap-based alignment improves on point-cloud fitting, since the loss is differentiable in vertex positions.
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

3 major / 5 minor

Summary. The paper proposes MGIoU, a loss function for optimizing convex parametric shapes that projects both shapes onto a set of unique normals and averages one-dimensional GIoU values. Three variants are introduced: MGIoU for structured convex shapes, MGIoU+ for unstructured convex shapes via a convexity regularizer, and MGIoU− for minimizing overlap in trajectory prediction. The authors claim that MGIoU unifies position, size, and orientation optimization in a single differentiable objective, is strongly correlated with IoU, satisfies metric properties and scale invariance, and reduces loss-computation latency by 10–40x. Experiments are reported on DOTAv1.5 for oriented object detection, Omni3D for 3D 6-DoF detection, ICDAR2017 for quadrilateral detection, and Waymo for trajectory prediction, with code made publicly available.

Significance. If the claims are substantiated, MGIoU would be a practically valuable unification: one loss replacing task-specific IoU approximations across detection and trajectory tasks, with sizable speedups and consistent empirical gains on standard benchmarks. The empirical results are credible as reported, and the code availability strengthens reproducibility. However, the theoretical properties that the paper emphasizes—especially identity and the triangle inequality—rest on unproven or incorrectly scoped arguments, and the central 'strongly correlated with IoU' assertion is not quantified. These gaps do not necessarily invalidate the empirical contribution, but they need to be fixed or substantially qualified before the paper's stronger claims can be accepted.

major comments (3)
  1. [Appendix Sec. 8, Lemma 2] The converse direction of Lemma 2 is asserted rather than proved: from equality of all 1D GIoU values (each equal to 1) the proof concludes P=G 'given that A includes all unique normals.' No argument or reference establishes that a finite set of edge/face normals determines a convex body, and the issue is especially delicate because A itself depends on the two shapes being compared. For rectangles and cuboids the claim may be true, but it must be demonstrated explicitly for the stated parametric families; without this, the identity property and the resulting metric claim are unproven.
  2. [Appendix Sec. 8, Proposition 1 (Triangle Inequality)] The triangle-inequality proof explicitly assumes 'a fixed set of projection directions A for all pairs of shapes under consideration,' but the definition of MGIoU in Section 3.2 makes A depend on the pair (P,G). For a triplet of differently oriented rectangles, A(P,R), A(P,Q), and A(Q,R) are generally different. The averaging argument therefore does not apply to the actual MGIoU definition. Additionally, the assertion that 1−GIoU1D satisfies the triangle inequality and 'inherits' it from [27] is not demonstrated; the cited reference treats GIoU for boxes, not this particular interval distance. This is a load-bearing gap because the paper advertises metric properties as a theoretical guarantee.
  3. [Section 1 and Section 3.2] The abstract and introduction claim that MGIoU is 'strongly correlated with IoU,' but no experiment or theorem quantifies this correlation. Since MGIoU is an average of 1D GIoU values over a small set of normals, it can rank pairs differently from true IoU in simple cases, e.g., two thin rectangles crossing at a small angle can produce large interval overlaps along their own axes while the true intersection area is small. The paper's aggregate mAP improvements do not directly validate the correlation claim. Please add a controlled study on random valid shape pairs reporting Pearson/Spearman correlation between MGIoU (and its variants) and true IoU, with the shape families used in the experiments, or explicitly qualify the claim.
minor comments (5)
  1. [Section 4, heading] 'Monocular 3D 6-DoF Object Regconition' contains a typo; it should be 'Recognition'.
  2. [Table 1 and Section 5.2] The baseline name 'Distangled-L1+Chamfer' appears in the table and text; if this is intended to be 'Disentangled-L1+Chamfer', the spelling should be corrected for consistency with the cited work.
  3. [Figure 1 caption] The caption contains a formatting error: 'MGIoU+[A) unstructured quadrilaterals' should be 'MGIoU+ (A) unstructured quadrilaterals'.
  4. [Section 5.1 and Table 2] The statement that MGIoU latency is 'only slightly higher than L1' is misleading given the table shows 0.45 ms vs 0.03 ms (a 15x difference), though the absolute difference is small; please rephrase to avoid overstating speed, e.g., by noting both are below 0.5 ms.
  5. [Section 4, trajectory prediction setting] The Waymo experiment is trained on a 20% subset of the training set, so the absolute numbers in Table 3 are not directly comparable to full-benchmark results; this limitation should be stated more prominently in the main text, not only near the table.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: MGIoU is a constructive loss defined before benchmark evaluation; the only self-citation (GIoU) is an external building block, not a fitted target.

full rationale

The derivation of MGIoU is not circular. MGIoU is defined in Sec. 3.2 and Algorithm 1 as an average of one-dimensional GIoU values over the unique normals of the predicted and ground-truth shapes; this is a constructive definition, not a quantity fitted to the reported benchmarks. The empirical claims (DOTAv1.5 mAP 0.554, Omni3D AP3D improvements, ICDAR2017 AP/AR, Waymo collision reduction of 14%) are obtained by training standard baselines with the proposed loss and comparing against existing losses; no loss parameter is fitted to those test sets. The only self-citation is to GIoU [27], coauthored by one of the present authors, which is used as a building block and is an independent, externally established prior result; using it does not make the new loss's predictions equivalent to its inputs. The appendix contains unproven geometric assertions — the converse of the identity lemma assumes that equality of finitely many 1D projections implies equality of the convex shapes, and the triangle-inequality proof assumes a fixed projection set A for all pairs even though A is pair-dependent — but these are rigor gaps, not circular reductions: the loss is not defined in terms of its claimed outputs, and the central experimental comparisons are benchmark-tested. Therefore no circularity is established.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

MGIoU introduces no new physical entities or particles. The free parameter lambda is introduced for the unstructured case. The key axioms are geometric: that the finite normal set preserves shape identity, that 1D GIoU distance is a metric, and that the average of 1D GIoU values tracks true overlap.

free parameters (1)
  • lambda (convexity weight)
    Added in LMGIoU+ to balance the convexity regularizer for unstructured shapes. The paper does not report the value used in the ICDAR2017 experiments.
assumptions (3)
  • domain assumption Unique normals of P and G (edge normals, face normals, semi-axes) determine each structured convex shape.
    Used in Section 3.1 to reduce shape overlap to 1D projections and in Lemma 2 of the Appendix to conclude that equality of all 1D projections implies shape equality.
  • domain assumption 1 - GIoU in one dimension satisfies the triangle inequality.
    Appendix Sec. 8 asserts this and credits GIoU [27], but no proof is given for intervals, and the proof that follows assumes a fixed direction set A for all shape pairs.
  • domain assumption Averaged per-normal GIoU correlates strongly with true shape IoU.
    This is the core proxy assumption that makes MGIoU a useful loss. The paper asserts it in the abstract and intro but does not provide a correlation analysis against exact IoU.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Marginalized Generalized IoU (MGIoU): A Unified Objective Function for Optimizing Any Convex Parametric Shapes." pith.science (2026). https://pith.science/paper/EEPZKFND

@misc{pith2026250416443,
  author       = {Pith},
  title        = {Pith review of: Marginalized Generalized IoU (MGIoU): A Unified Objective Function for Optimizing Any Convex Parametric Shapes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EEPZKFND}},
  note         = {Machine review of arXiv:2504.16443}
}
read the original abstract

Optimizing the similarity between parametric shapes is crucial for numerous computer vision tasks, where Intersection over Union (IoU) stands as the canonical measure. However, existing optimization methods exhibit significant shortcomings: regression-based losses like L1/L2 lack correlation with IoU, IoU-based losses are unstable and limited to simple shapes, and task-specific methods are computationally intensive and not generalizable accross domains. As a result, the current landscape of parametric shape objective functions has become scattered, with each domain proposing distinct IoU approximations. To address this, we unify the parametric shape optimization objective functions by introducing Marginalized Generalized IoU (MGIoU), a novel loss function that overcomes these challenges by projecting structured convex shapes onto their unique shape Normals to compute one-dimensional normalized GIoU. MGIoU offers a simple, efficient, fully differentiable approximation strongly correlated with IoU. We then extend MGIoU to MGIoU+ that supports optimizing unstructured convex shapes. Together, MGIoU and MGIoU+ unify parametric shape optimization across diverse applications. Experiments on standard benchmarks demonstrate that MGIoU and MGIoU+ consistently outperform existing losses while reducing loss computation latency by 10-40x. Additionally, MGIoU and MGIoU+ satisfy metric properties and scale-invariance, ensuring robustness as an objective function. We further propose MGIoU- for minimizing overlaps in tasks like collision-free trajectory prediction. Code is available at https://ldtho.github.io/MGIoU

Figures

Figures reproduced from arXiv: 2504.16443 by the authors.

Figure 1
Figure 1. MGIoU and its variants computation. Predicted ( [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Qualitative visualisation (Test set images) comparing [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Qualitative visualisation on Omni3D dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Qualitative visualisation on Waymo dataset, we visualize the predicted future bounding boxes of road agents in the next 8 seconds [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. LeAD-M3D: Leveraging Asymmetric Distillation for Real-Time Monocular 3D Detection

    cs.CV 2025-12 conditional novelty 6.0 of 10

    LeAD-M3D reports state-of-the-art monocular 3D detection accuracy on KITTI, Waymo, and Rope3D at real-time speeds using mixup-based denoising distillation, 3D-aware matching, and confidence-gated inference.

Reference graph

Works this paper leans on

53 extracted references · 35 canonical work pages · cited by 1 Pith paper

  1. [27]

    Generalized in- tersection over union: A metric and a loss for bounding box regression

    Hamid Rezatofighi, Nathan Tsoi, JunYoung Gwak, Amir Sadeghian, Ian Reid, and Silvio Savarese. Generalized in- tersection over union: A metric and a loss for bounding box regression. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 658–666,

  2. [1]

    Super-gradients, 2021

    Shay Aharon, Louis-Dupont, Ofri Masad, Kate Yurkova, Lotem Fridman, Lkdci, Eugene Khvedchenya, Ran Rubin, Natan Bagrov, Borys Tymchenko, Tomer Keren, Alexander Zhilko, and Eran-Deci. Super-gradients, 2021. 6, 7

    work page 2021

  3. [2]

    Objectron: A large scale dataset of object-centric videos in the wild with pose an- notations

    Adel Ahmadyan, Liangkai Zhang, Artsiom Ablavatski, Jian- ing Wei, and Matthias Grundmann. Objectron: A large scale dataset of object-centric videos in the wild with pose an- notations. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7822–7831,

  4. [3]

    Arkitscenes: A diverse real-world dataset for 3d indoor scene understanding using mobile rgb-d data

    Gilad Baruch, Zhuoyuan Chen, Afshin Dehghan, Tal Dimry, Yuri Feigin, Peter Fu, Thomas Gebauer, Brandon Joffe, Daniel Kurz, Arik Schwartz, et al. Arkitscenes: A diverse real-world dataset for 3d indoor scene understanding using mobile rgb-d data. arXiv preprint arXiv:2111.08897, 2021. 3, 6

    arXiv 2021

  5. [4]

    Disentangled contour learn- ing for quadrilateral text detection

    Yanguang Bi and Zhiqiang Hu. Disentangled contour learn- ing for quadrilateral text detection. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 909–918, 2021. 2

    work page 2021

  6. [5]

    Omni3D: A large benchmark and model for 3D object detection in the wild

    Garrick Brazil, Abhinav Kumar, Julian Straub, Nikhila Ravi, Justin Johnson, and Georgia Gkioxari. Omni3D: A large benchmark and model for 3D object detection in the wild. In CVPR, Vancouver, Canada, 2023. IEEE. 2, 3, 6, 7

    work page 2023

  7. [6]

    nuscenes: A multi- modal dataset for autonomous driving

    Holger Caesar, Varun Bankiti, Alex H Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Gi- ancarlo Baldan, and Oscar Beijbom. nuscenes: A multi- modal dataset for autonomous driving. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 11621–11631, 2020. 6

    2020

  8. [7]

    Shapenet: An information-rich 3d model repository

    Angel X Chang, Thomas Funkhouser, Leonidas Guibas, Pat Hanrahan, Qixing Huang, Zimo Li, Silvio Savarese, Manolis Savva, Shuran Song, Hao Su, et al. Shapenet: An information-rich 3d model repository. arXiv preprint arXiv:1512.03012, 2015. 3

    arXiv 2015

Show all 53 references
  1. [8]

    MMCV: OpenMMLab computer vision foundation

    MMCV Contributors. MMCV: OpenMMLab computer vision foundation. https : / / github . com / open - mmlab/mmcv, 2018. 2

    2018

  2. [9]

    Scannet: Richly-annotated 3d reconstructions of indoor scenes

    Angela Dai, Angel X Chang, Manolis Savva, Maciej Hal- ber, Thomas Funkhouser, and Matthias Nießner. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 5828–5839, 2017. 3

    2017

  3. [10]

    Vision meets robotics: The kitti dataset

    Andreas Geiger, Philip Lenz, Christoph Stiller, and Raquel Urtasun. Vision meets robotics: The kitti dataset. The in- ternational journal of robotics research, 32(11):1231–1237,

  4. [11]

    Icdar2017 robust reading challenge on coco-text

    Raul Gomez, Baoguang Shi, Lluis Gomez, Lukas Numann, Andreas Veit, Jiri Matas, Serge Belongie, and Dimosthenis Karatzas. Icdar2017 robust reading challenge on coco-text. In 2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), pages 1435–1443. IEEE,

    2017

  5. [12]

    An end-to-end quadrilateral regression network for comic panel extraction

    Zheqi He, Yafeng Zhou, Yongtao Wang, Siwei Wang, Xiao- qing Lu, Zhi Tang, and Ling Cai. An end-to-end quadrilateral regression network for comic panel extraction. In Proceed- ings of the 26th ACM international conference on Multime- dia, pages 887–895, 2018. 2, 3, 8

    2018

  6. [13]

    Quad- box: A new approach for arbitrary quadrilateral detection

    Pranav Keserwani, Rajiv Singh, and Rajesh Shukla. Quad- box: A new approach for arbitrary quadrilateral detection. Neurocomputing, 448:188–200, 2021. 3

    2021

  7. [14]

    Diverse multiple trajectory prediction using a two-stage prediction network trained with lane loss

    Sanmin Kim, Hyeongseok Jeon, Jun Won Choi, and Dong- suk Kum. Diverse multiple trajectory prediction using a two-stage prediction network trained with lane loss. IEEE Robotics and Automation Letters, 8(4):2038–2045, 2022. 3

    2022

  8. [15]

    Unimode: Unified monocular 3d object detection

    Zhuoling Li, Xiaogang Xu, SerNam Lim, and Hengshuang Zhao. Unimode: Unified monocular 3d object detection. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 16561–16570, 2024. 2

    2024

  9. [16]

    Textboxes++: A single-shot oriented scene text detector.IEEE transactions on image processing, 27(8):3676–3690, 2018

    Minghui Liao, Baoguang Shi, and Xiang Bai. Textboxes++: A single-shot oriented scene text detector.IEEE transactions on image processing, 27(8):3676–3690, 2018. 2, 3

    2018

  10. [17]

    Focal loss for dense object detection

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Doll´ar. Focal loss for dense object detection. In Pro- ceedings of the IEEE international conference on computer vision, pages 2980–2988, 2017. 6

    2017

  11. [18]

    Deep matching prior network: Toward tighter multi-oriented text detection

    Yuliang Liu and Lianwen Jin. Deep matching prior network: Toward tighter multi-oriented text detection. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 1962–1969, 2017. 2

    1962

  12. [19]

    Yolo-pose: Enhancing yolo for multi person pose estimation using object keypoint similarity loss

    Debapriya Maji, Soyeb Nagori, Manu Mathew, and Deepak Poddar. Yolo-pose: Enhancing yolo for multi person pose estimation using object keypoint similarity loss. In Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2637–2646, 2022. 2, 8

    2022

  13. [20]

    Improv- ing movement prediction of traffic actors using off-road loss and bias mitigation

    Matthew Niedoba, Henggang Cui, Kevin Luo, Darshan Hegde, Fang-Chieh Chou, and Nemanja Djuric. Improv- ing movement prediction of traffic actors using off-road loss and bias mitigation. In Workshop on’Machine Learning for Autonomous Driving’at Conference on Neural Information P...

    2019

  14. [21]

    Pointnet: Deep learning on point sets for 3d classification and segmentation

    Charles R Qi, Hao Su, Kaichun Mo, and Leonidas J Guibas. Pointnet: Deep learning on point sets for 3d classification and segmentation. arXiv preprint arXiv:1612.00593, 2017. 2, 3

    2017 arXiv

  15. [22]

    Learning modulated loss for rotated object detection

    Wen Qian, Xue Yang, Silong Peng, Junchi Yan, and Yue Guo. Learning modulated loss for rotated object detection. In Proceedings of the AAAI conference on artificial intelli- gence, pages 2458–2466, 2021. 3

    2021

  16. [23]

    Accelerating 3d deep learning with pytorch3d

    Nikhila Ravi, Jeremy Reizenstein, David Novotny, Tay- lor Gordon, Wan-Yen Lo, Justin Johnson, and Georgia Gkioxari. Accelerating 3d deep learning with pytorch3d. arXiv:2007.08501, 2020. 2

    2007 arXiv

  17. [24]

    Yolo9000: better, faster, stronger

    Joseph Redmon and Ali Farhadi. Yolo9000: better, faster, stronger. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 7263–7271, 2017. 3

    2017

  18. [25]

    You only look once: Unified, real-time object de- tection

    Joseph Redmon, Santosh Divvala, Ross Girshick, and Ali Farhadi. You only look once: Unified, real-time object de- tection. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 779–788, 2016. 2

    2016

  19. [26]

    Faster r-cnn: Towards real-time object detection with region 9 proposal networks

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: Towards real-time object detection with region 9 proposal networks. Advances in neural information process- ing systems, 28, 2015. 3

    2015

  20. [28]

    Hypersim: A photorealistic syn- thetic dataset for holistic indoor scene understanding

    Mike Roberts, Jason Ramapuram, Anurag Ranjan, Atulit Kumar, Miguel Angel Bautista, Nathan Paczan, Russ Webb, and Joshua M Susskind. Hypersim: A photorealistic syn- thetic dataset for holistic indoor scene understanding. In Proceedings of the IEEE/CVF international conference o...

    2021

  21. [29]

    Motion transformer with global intention localization and lo- cal movement refinement

    Shaoshuai Shi, Li Jiang, Dengxin Dai, and Bernt Schiele. Motion transformer with global intention localization and lo- cal movement refinement. Advances in Neural Information Processing Systems, 2022. 2, 6

    2022

  22. [30]

    Mtr++: Multi-agent motion prediction with symmetric scene modeling and guided intention querying

    Shaoshuai Shi, Li Jiang, Dengxin Dai, and Bernt Schiele. Mtr++: Multi-agent motion prediction with symmetric scene modeling and guided intention querying. arXiv preprint arXiv:2306.17770, 2023. 2

    2023 arXiv

  23. [31]

    Sun rgb-d: A rgb-d scene understanding benchmark suite

    Shuran Song, Samuel P Lichtenberg, and Jianxiong Xiao. Sun rgb-d: A rgb-d scene understanding benchmark suite. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 567–576, 2015. 6

    2015

  24. [32]

    Deep learning on the ro- tation manifold for object detection in 3d point clouds.IEEE Transactions on Pattern Analysis and Machine Intelligence,

    Hang Su, Varun Jampani, Deqing Sun, Subhransu Maji, Vladlen Koltun, and Philip Torr. Deep learning on the ro- tation manifold for object detection in 3d point clouds.IEEE Transactions on Pattern Analysis and Machine Intelligence,

  25. [33]

    Scalability in perception for autonomous driving: Waymo open dataset

    Pei Sun, Henrik Kretzschmar, Xerxes Dotiwalla, Aurelien Chouard, Vijaysai Patnaik, Paul Tsui, James Guo, Yin Zhou, Yuning Chai, Benjamin Caine, et al. Scalability in perception for autonomous driving: Waymo open dataset. In Proceed- ings of the IEEE/CVF conference on computer ...

    2020

  26. [34]

    Trafficsim: Learning to simulate realistic multi- agent behaviors

    Simon Suo, Sebastian Regalado, Sergio Casas, and Raquel Urtasun. Trafficsim: Learning to simulate realistic multi- agent behaviors. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition , pages 10400–10409, 2021. 3

    2021

  27. [35]

    Dota: A large-scale dataset for object detection in aerial images

    Gui-Song Xia, Xu Bai, Jian Ding, Zhen Shen, Zengpu Ma, Hao Wang, Jianguo Gao, Dandan Zhang, Ming Li, Baogang He, et al. Dota: A large-scale dataset for object detection in aerial images. arXiv preprint arXiv:1803.06165, 2018. 6

    2018 arXiv

  28. [36]

    R3det: Refined joint representation for oriented object detection

    Xiaolong Yang, Qijie Liu, Jianwu Yan, and Anran Li. R3det: Refined joint representation for oriented object detection. arXiv preprint arXiv:2105.02445, 2021. 2, 3

    2021 arXiv

  29. [37]

    Rethinking rotated object detection with gaussian wasserstein distance loss

    Xue Yang, Junchi Yan, Qi Ming, Wentao Wang, Xiaopeng Zhang, and Qi Tian. Rethinking rotated object detection with gaussian wasserstein distance loss. In International con- ference on machine learning , pages 11830–11841. PMLR,

  30. [38]

    Learning high-precision bounding box for rotated object detection via kullback- leibler divergence

    Xue Yang, Xiaojiang Yang, Jirui Yang, Qi Ming, Wentao Wang, Qi Tian, and Junchi Yan. Learning high-precision bounding box for rotated object detection via kullback- leibler divergence. Advances in Neural Information Process- ing Systems, 34:18381–18394, 2021. 2, 3, 6, 7

    2021

  31. [39]

    The kfiou loss for rotated object detection

    Xue Yang, Yue Zhou, Gefan Zhang, Jirui Yang, Wentao Wang, Junchi Yan, Xiaopeng Zhang, and Qi Tian. The kfiou loss for rotated object detection. arXiv preprint arXiv:2201.12558, 2022. 2, 3, 6, 7

    2022 arXiv

  32. [40]

    Trajgen: Generating realistic and diverse trajectories with re- active and feasible agent behaviors for autonomous driving

    Qichao Zhang, Yinfeng Gao, Yikang Zhang, Youtian Guo, Dawei Ding, Yunpeng Wang, Peng Sun, and Dongbin Zhao. Trajgen: Generating realistic and diverse trajectories with re- active and feasible agent behaviors for autonomous driving. IEEE Transactions on Intelligent Transportati...

    2022

  33. [41]

    Distance-iou loss: Faster and bet- ter learning for bounding box regression

    Zhaohui Zheng, Ping Wang, Wei Liu, Jinze Li, Rongguang Ye, and Dongwei Ren. Distance-iou loss: Faster and bet- ter learning for bounding box regression. In Proceedings of the AAAI conference on artificial intelligence, pages 12993– 13000, 2020. 3

    2020

  34. [42]

    Enhancing ge- ometric factors in model learning and inference for object detection and instance segmentation

    Zhaohui Zheng, Ping Wang, Dongwei Ren, Wei Liu, Rong- guang Ye, Qinghua Hu, and Wangmeng Zuo. Enhancing ge- ometric factors in model learning and inference for object detection and instance segmentation. IEEE Transactions on cybernetics, 52(8):8574–8586, 2021. 3

    2021

  35. [43]

    Iou loss for 2d/3d ob- ject detection

    Dingfu Zhou, Jin Fang, Xibin Song, Chenye Guan, Junbo Yin, Yuchao Dai, and Ruigang Yang. Iou loss for 2d/3d ob- ject detection. In 2019 international conference on 3D vision (3DV), pages 85–94. IEEE, 2019. 2

    2019

  36. [44]

    Mmrotate: A rotated object detection benchmark using pytorch

    Yue Zhou, Xue Yang, Gefan Zhang, Jiabao Wang, Yanyi Liu, Liping Hou, Xue Jiang, Xingzhao Liu, Junchi Yan, Chengqi Lyu, Wenwei Zhang, and Kai Chen. Mmrotate: A rotated object detection benchmark using pytorch. In Proceedings of the 30th ACM International Conference on Multimedia ,

  37. [46]

    Simplification of One-Dimensional GIoU Formula We aim to prove that the simplified formula for the one- dimensional Generalized Intersection over Union (GIoU) metric for projections onto axisAi is equivalent to the stan- dard GIoU definition. The simplified formula that is use...

  38. [47]

    Specifically, we analyze the following properties forLMGIoU over structured convex shapesP andG with a shared para- metric domain:

    Proof of Properties for MGIoU To assess the robustness of the MGIoU as a similarity mea- sure for shape optimization, we define the loss function LMGIoU(P,G ) = 1−MGIoU(P,G) 2 and evaluate its properties. Specifically, we analyze the following properties forLMGIoU over structu...

  39. [52]

    Scale-Invariance: LMGIoU(sP,sG ) = LMGIoU(P,G ) for any scalars> 0. Definition of MGIoU MGIoU is defined as the average of one-dimensional Gen- eralized Intersection over Union (GIoU) values computed across a set of projection directions: MGIoU(P,G ) = 1 |A| X i∈A GIoU1D i (Pi...

  40. [53]

    Non-negativity:LMGIoU(P,G )≥ 0,

  41. [54]

    Identity:LMGIoU(P,G ) = 0 if and only ifP =G,

  42. [55]

    Symmetry:LMGIoU(P,G ) =LMGIoU(G,P ),

  43. [56]

    Triangle Inequality:LMGIoU(P,R )≤L MGIoU(P,Q ) + LMGIoU(Q,R ),

  44. [57]

    Scale-Invariance:LMGIoU(sP,sG ) =LMGIoU(P,G ) for any scalars> 0. Proof 4 - Non-negativity: Since MGIoU (P,G ) ≤ 1, we have: LMGIoU(P,G ) = 1− MGIoU(P,G ) 2 ≥ 0 - Identity: From Lemma 2,LMGIoU(P,G ) = 0 if and only if MGIoU(P,G ) = 1 , which holds if and only ifP =G. - Symmetr...

  45. [2022]

    6 10 Marginalized Generalized IoU (MGIoU): A Unified Objective Function for Optimizing Any Convex Parametric Shapes Supplementary Material

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.