Pith. sign in

REVIEW 4 major objections 6 minor 31 references

MonoDINO-DETR: Depth-Enhanced Monocular 3D Object Detection Using a Vision Foundation Model

T0 review · 4 major / 6 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read MonoDINO-DETR claims that a camera-only detection transformer using a DINOv2 vision foundation model backbone outperforms recent monocular 3D detectors on KITTI and on a high-bank racing dataset, without needing LiDAR or ground-plane…

desk verdict Useful engineering combination with a plausible KITTI gain, but the custom-dataset comparison is confounded by protocol asymmetry and unvalidated pseudo-labels. read the letter →

arxiv 2502.00315 v1 pith:5DS65U7C submitted 2025-02-01 cs.CV

Jihyeok Kim , Seongwoo Moon , Sungwon Nah , David Hyunchul Shim This is my paper · ORCID
classification cs.CV
keywords monocular3DobjectdetectionvisionfoundationmodelDINOv2DETRdepthestimationKITTIbenchmarkautonomousracingdynamicanchorboxes
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 tries to show that a self-supervised vision foundation model can replace CNN backbones in monocular 3D object detection and improve both depth estimation and localization. The proposed detector, MonoDINO-DETR, uses a DINOv2 Vision Transformer as backbone, fuses features across layers with transposed convolutions to recover multi-scale detail, and borrows a large pretrained relative-depth decoder for depth. On KITTI's car benchmark it reports the best validation AP3D and APBEV among the cited methods, up to 27.93 easy / 19.39 moderate / 15.97 hard AP3D with dynamic anchor boxes, without relying on LiDAR or ground-plane assumptions. On a self-collected Kentucky Speedway racing dataset it reports 26.23 AP3D at IoU 0.7, far above MonoDETR's 9.86, suggesting strong transfer to high-elevation environments. If these results hold, camera-only perception becomes a more credible fallback for autonomous racing and urban driving.

What carries the argument

The core mechanism is the pairing of a self-supervised Vision Transformer foundation model with two adapters: a Hierarchical Feature Fusion Block that builds multi-resolution feature maps from DINOv2's single-scale layers using transposed convolutions, and a Dense Prediction Transformer head initialized from Depth Anything V2 for pixel-wise depth. Depth is supervised only through object-wise depth bins created with linear increasing discretization, so no dense depth labels are required. In the decoder, 6D dynamic anchor queries, encoding a reference point and distances to the left, right, top, and bottom edges, are refined layer by layer to improve localization of asymmetric objects.

What would settle it

Re-evaluate the racing dataset with human-verified or sensor-fused 3D labels and recompute AP3D; if the IoU 0.7 gap between MonoDINO-DETR and MonoDETR shrinks from 16.4 points to within label noise, the transfer claim would be undermined.

Watch

Extended reading notes

Core claim

The central claim is that a plain, non-hierarchical Vision Transformer foundation model, DINOv2, can serve as the visual backbone for one-stage monocular 3D detection, provided its single-scale features are reorganized into a hierarchy and its depth stream is initialized from a large-scale relative-depth model. The paper argues that DINOv2's global context improves depth estimation compared with CNN backbones, that the Hierarchical Feature Fusion Block recovers the multi-scale local cues needed for localization, and that the transformer decoder's 6D dynamic anchor boxes refine asymmetric box shapes. Together these components let MonoDINO-DETR outperform recent monocular detectors on KITTI validation for the car class, including methods that require LiDAR or ground-plane information, and produce large gains on a racetrack dataset with elevation changes. The authors state that the model operates end-to-end in a one-stage manner and needs no extra data beyond images and object-wise depth supervision.

Load-bearing premise

The large custom-dataset gains rest on PointPillars-generated pseudo-labels being accurate enough to serve as ground truth for comparing two monocular detectors on a high-bank racetrack.

Editorial extensions

If this is right

  • On KITTI validation for the car class, MonoDINO-DETR plus dynamic anchor boxes reaches 27.93 / 19.39 / 15.97 AP3D for easy, moderate, and hard difficulty, surpassing all cited detectors, including those that use LiDAR or ground-plane priors, without extra data.
  • The ablation of the Hierarchical Feature Fusion Block attributes roughly 3.1 / 2.5 / 2.2 AP3D points of easy, moderate, and hard improvement over the best alternative fusion design, making multi-layer, multi-resolution fusion load-bearing for the reported result.
  • The 6D dynamic anchor boxes add about 1.2 AP3D points on easy difficulty and smaller gains on moderate and hard difficulty, so the refinement mainly helps closer, easier objects.
  • On the custom racing dataset, the reported 26.23 versus 9.86 AP3D at IoU 0.7 relative to MonoDETR implies the method transfers to high-bank tracks where flat-ground assumptions fail.
  • At 66 to 74 milliseconds of inference time on the reported hardware, the method remains within a single-stage real-time operating budget.

Reading between the lines

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

  • A likely reason DINOv2 helps is that its pretraining on massive unlabeled images transfers general notions of object layout and scale; a testable corollary is that other self-supervised Vision Transformers would give similar gains when wrapped in the same fusion and depth-decoder setup.
  • Because the custom dataset's labels are generated by PointPillars, part of the 16-point AP3D gain over MonoDETR could be label-noise bias in favor of the better detector; manually auditing a subset of boxes would separate detector quality from label quality.
  • If the depth-transfer component is the key ingredient, then adding the same depth module to other DETR-style monocular detectors should lift them by roughly the amount seen in the first ablation, about 1.2 AP3D points on moderate difficulty.
  • The method's explicit avoidance of ground-plane priors suggests it should be tested on datasets with road slope or elevation changes beyond KITTI, where plane-assumption methods are known to degrade.
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 paper proposes MonoDINO-DETR, a one-stage monocular 3D object detector that replaces CNN backbones with a DINOv2 Vision Transformer, uses a DPT head initialized from Depth Anything V2 for depth features, introduces a Hierarchical Feature Fusion Block (HFFB) to create multi-scale visual features, and extends DAB-DETR's dynamic anchor queries to six dimensions. Training uses object-wise depth supervision with LID depth bins and a combination of 2D/3D losses. On the KITTI validation split for the car class, the paper reports AP3D of 27.93/19.39/15.97 (easy/moderate/hard) with DAB, and claims to outperform all recent state-of-the-art methods without using additional data. A custom racing dataset of 1,171 training and 293 validation frames, pseudo-labeled with PointPillars, is used to claim large gains over MonoGround and MonoDETR and to argue for the superior generalizability of foundation models.

Significance. If the KITTI validation numbers are reproducible, the improvements are meaningful and would provide evidence that ViT-based foundation models can benefit monocular 3D detection. The paper's strengths include a clear architecture description, released code, and ablations isolating the contributions of HFFB and 6D dynamic anchors. The significance is currently tempered by three issues: the custom-dataset comparison does not specify whether the baselines were trained on the custom split, the 'no additional data' claim obscures the large external pretraining corpora used by DINOv2 and Depth Anything V2, and the KITTI headline rests on single-run validation metrics without error bars or official test-set confirmation. These issues are fixable but need to be addressed before the broader claims can be accepted.

major comments (4)
  1. [IV-B, Table II] The custom-dataset comparison is uninterpretable without stating whether MonoGround and MonoDETR were trained or fine-tuned on the 1,171 custom training frames described in Section IV-A. The text says only that these models were 'tested' and that they 'do not require extra data.' If they were applied off-the-shelf from KITTI while MonoDINO-DETR was trained or fine-tuned on the custom split, the reported margins (AP3D 26.23 vs 9.86 vs 1.49 at IoU 0.7) measure adaptation advantage, not generalizability. The paper should specify the exact training protocol for every method in Table II and rerun the comparison with matched training data, schedule, and evaluation protocol.
  2. [IV-A] The custom dataset uses PointPillars pseudo-labels as ground truth for both training and evaluation, but no analysis of label quality is provided. Because the track has elevation changes and LiDAR sparsity, systematic errors in pseudo-labels could inflate or deflate AP in ways that differ across detectors. The paper should report label-quality statistics, such as agreement with manually annotated boxes or precision/recall as a function of distance and elevation, and ideally evaluate on a manually annotated subset to support the custom-dataset claims.
  3. [Abstract, Table I] The claim that the method works 'without requiring any additional data' is overstated. The method depends on DINOv2 pretrained on LVD-142M and on Depth Anything V2 pretrained on large-scale synthetic data. If 'additional data' means training-time supervision from LiDAR or depth maps, the paper should define it that way and explicitly acknowledge that the comparison with ImageNet-pretrained baselines is not a comparison of total data used. As written, the claim invites the reading that the method uses less data than prior work, which is not established by the paper.
  4. [IV-B, Table I] The KITTI state-of-the-art claim rests on single-run validation AP values at 40 recall positions, with no error bars and no official test-set confirmation, while the implementation details show hyperparameters (learning rate schedule, depth range, number of bins) were selected on this same validation split. The authors should either report multiple seeds or official test-server results, or soften the 'outperforms all recent models' claim to acknowledge the limitation of a single validation run.
minor comments (6)
  1. [IV-A] The word 'splited' should be 'split', and the KITTI split used should be cited precisely (the paper cites Chen et al. but should state that the 3,712/3,769 split is the standard training/validation split).
  2. [Table II] Table II lists 'MonoDINO-DETR (small)' but the implementation details do not define the small variant; please specify the backbone size, depth, and training settings used for the small model.
  3. [V] The conclusion calls the method a 'real-time framework', but Table I reports 74 ms inference time, which is below 30 FPS; please define the real-time threshold used or remove the term.
  4. [III-A] The notation f^3_{1/14} and the description that transposed convolutions produce features '4, 2, and 1 times larger than the original' is confusing; please clarify the downsampling ratios and how the 1/14 notation relates to the input resolution.
  5. [Table IV] The 'Improvement v.s. second-best' row in Table IV should state explicitly that the comparison is against the best HFFB variant, not against a fixed external baseline, to avoid ambiguity.
  6. [Table V] The text says the DAB variant outperforms 'across most metrics and difficulty levels', but the moderate and hard AP3D gains are small (19.19 to 19.39 and 15.92 to 15.97); please report the effect size and note the marginal nature of these improvements.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the performance claims are empirically evaluated benchmark results, not quantities derived from their own inputs.

full rationale

This is an empirical system paper. The central claims are benchmark metrics (AP3D and APBEV on KITTI val and a custom racing dataset), and these metrics are obtained by training the proposed architecture and measuring detection accuracy against externally defined labels. No equation in the paper reduces a predicted quantity to a fitted input. The model borrows DINOv2, DPT, Depth Anything V2, MonoDETR, and DAB-DETR components, but these are external building blocks, and the ablations compare variants of the proposed method against each other and against existing published results. There is no self-citation chain, no uniqueness theorem imported from the authors' prior work, and no parameter fitted to a subset and then reported as a prediction of the same quantity. The custom dataset comparison may raise a methodological concern about whether the baselines were trained on the same 1,171 custom frames, because the text says only that MonoGround and MonoDETR were 'tested' on the custom dataset; however, that concern is about experimental fairness and validity, not circularity. Even if the custom-dataset protocol were asymmetric, the claim would be under-supported rather than circular-by-construction. The KITTI result is a standard held-out validation evaluation on the canonical split and is not equivalent to the training objective by definition. Therefore no significant circularity is present.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The main ledger items are the hand-chosen depth ranges/bin counts and the domain assumptions about transfer learning and pseudo-label quality. There are no invented physical entities; the 6D anchor is a representation choice. The biggest uncharged item is the reliance on external pretraining corpora, which is not counted as a free parameter but is a major reproducibility and fairness factor.

free parameters (5)
  • Depth bin count k = 80 (KITTI), 160 (custom)
    Chosen by hand for each dataset in Section IV-A; affects the granularity of depth-map supervision.
  • Depth range [dmin,dmax] = [0,60] m KITTI, [0,120] m custom
    Assumed maximum detectable depth; set per dataset in Section IV-A.
  • Focal loss factors alpha and gamma = not reported
    Equation (1) defines alpha in [0,1] and gamma in [0,5] but specific values are omitted, so replication requires guessing.
  • Learning rate schedule = 0.0002, decay x0.1 at epochs 125 and 165, batch 8, 4 GPUs
    Standard training hyperparameters reported in Section IV-A; central performance depends on them but they are not novel.
  • HFFB transposed convolution geometry = not specified
    The multi-scale feature fusion design in Section III-A is described only qualitatively; kernel sizes, strides, and channel counts are free design choices.
assumptions (3)
  • domain assumption Monocular depth can be estimated from RGB via a DINOv2 backbone and DPT head fine-tuned from Depth Anything V2 weights.
    Section III-A assumes the pretrained relative depth features transfer to absolute depth for 3D box regression.
  • ad hoc to paper Pseudo-labels generated by PointPillars on LiDAR are accurate enough to evaluate monocular 3D detectors.
    Section IV-A uses these labels as ground truth for the custom racing dataset without a validation of label quality.
  • domain assumption KITTI val AP at 40 recall positions is a sufficient proxy for benchmark-level comparison.
    Section IV-B uses only validation metrics, not the official test server, to claim state-of-the-art performance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MonoDINO-DETR: Depth-Enhanced Monocular 3D Object Detection Using a Vision Foundation Model." pith.science (2026). https://pith.science/paper/5DS65U7C

@misc{pith2026250200315,
  author       = {Pith},
  title        = {Pith review of: MonoDINO-DETR: Depth-Enhanced Monocular 3D Object Detection Using a Vision Foundation Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5DS65U7C}},
  note         = {Machine review of arXiv:2502.00315}
}
read the original abstract

This paper proposes novel methods to enhance the performance of monocular 3D object detection models by leveraging the generalized feature extraction capabilities of a vision foundation model. Unlike traditional CNN-based approaches, which often suffer from inaccurate depth estimation and rely on multi-stage object detection pipelines, this study employs a Vision Transformer (ViT)-based foundation model as the backbone, which excels at capturing global features for depth estimation. It integrates a detection transformer (DETR) architecture to improve both depth estimation and object detection performance in a one-stage manner. Specifically, a hierarchical feature fusion block is introduced to extract richer visual features from the foundation model, further enhancing feature extraction capabilities. Depth estimation accuracy is further improved by incorporating a relative depth estimation model trained on large-scale data and fine-tuning it through transfer learning. Additionally, the use of queries in the transformer's decoder, which consider reference points and the dimensions of 2D bounding boxes, enhances recognition performance. The proposed model outperforms recent state-of-the-art methods, as demonstrated through quantitative and qualitative evaluations on the KITTI 3D benchmark and a custom dataset collected from high-elevation racing environments. Code is available at https://github.com/JihyeokKim/MonoDINO-DETR.

Figures

Figures reproduced from arXiv: 2502.00315 by the authors.

Figure 1
Figure 1. Illustration of the racing track environment and a [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overall Structure of MonoDINO-DETR. The proposed method, MonoDINO-DETR, is composed of four main components: the Feature Extraction Module, the Object-Wise Supervision Module, the Depth-Aware Transformer, and the MLP-Based Detection Heads. The visual feature extraction process is represented in green, while the depth feature extraction process is represented in blue. III. METHODOLOGY The overall architecture of the … view at source ↗
Figure 3
Figure 3. Overall Structure of Feature Extraction Module. The Feature Extraction Module is divided into three components: the DINOv2 backbone, the Visual Feature Extraction Module, and the Depth Feature Extraction Module. The Hierarchical Feature Fusion Block serves as the key module for visual features, while the combination of the DPT Head and DINOv2, which together form the Depth Anything V2 architecture, serves as the key… view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: Indy Race Car Platform. Synchronized camera image data and LiDAR data were acquired during the race using a front-mounted Luminar Iris LiDAR sensor and a front-left Mako G-319C camera. bounding box labels in the KITTI format. Evaluation metrics. We report the detection…
Figure 6
Figure 6. Figure 6: Ablation Study 2: Effect of the Hierarchical Feature Fusion Block. We evaluated the effect of HFFB by testing 3 HFFB variants. effectively incorporates the spatial information of bounding boxes and refines object queries layer by layer, leading to improved performance …
Figure 7
Figure 7. Figure 7: Qualitative results on the KITTI val set for the car class. The proposed method (green) and ground truth (red) [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Qualitative results on the custom dataset. Comparison of detection results between the proposed model (blue), the state-of-the-art models (green), and ground truth (red) in ego-view (left) and bird’s-eye view (right); MonoDETR (left) and MonoGround (right). [18] M. Car…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 12 canonical work pages

  1. [1]

    End-to-end object detection with transformers,

    N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, and S. Zagoruyko, “End-to-end object detection with transformers,” in European conference on computer vision . Springer, 2020, pp. 213– 229

    2020

  2. [2]

    Dinov2: Learning robust visual features without supervision,

    M. Oquab, T. Darcet, T. Moutakanni, H. V o, M. Szafraniec, V . Khali- dov, P. Fernandez, D. Haziza, F. Massa, A. El-Nouby et al., “Dinov2: Learning robust visual features without supervision,” arXiv preprint arXiv:2304.07193, 2023

    arXiv 2023

  3. [3]

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

    A. Dosovitskiy, “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929 , 2020

    arXiv 2010

  4. [4]

    Are we ready for autonomous driving? the kitti vision benchmark suite,

    A. Geiger, P. Lenz, and R. Urtasun, “Are we ready for autonomous driving? the kitti vision benchmark suite,” in 2012 IEEE conference on computer vision and pattern recognition . IEEE, 2012, pp. 3354– 3361

    2012

  5. [5]

    M3d-rpn: Monocular 3d region proposal network for object detection,

    G. Brazil and X. Liu, “M3d-rpn: Monocular 3d region proposal network for object detection,” in Proceedings of the IEEE/CVF in- ternational conference on computer vision , 2019, pp. 9287–9296

    work page 2019

  6. [6]

    Monoground: Detecting monocular 3d objects from the ground,

    Z. Qin and X. Li, “Monoground: Detecting monocular 3d objects from the ground,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 3793–3802

    work page 2022

  7. [7]

    Learning depth-guided convolutions for monocular 3d object detection,

    M. Ding, Y . Huo, H. Yi, Z. Wang, J. Shi, Z. Lu, and P. Luo, “Learning depth-guided convolutions for monocular 3d object detection,” in Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition workshops , 2020, pp. 1000–1001

    work page 2020

  8. [8]

    Depth-conditioned dynamic message propagation for monocular 3d object detection,

    L. Wang, L. Du, X. Ye, Y . Fu, G. Guo, X. Xue, J. Feng, and L. Zhang, “Depth-conditioned dynamic message propagation for monocular 3d object detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021, pp. 454–463

    work page 2021

Show all 31 references
  1. [9]

    Monodtr: Monocular 3d object detection with depth-aware transformer,

    K.-C. Huang, T.-H. Wu, H.-T. Su, and W. H. Hsu, “Monodtr: Monocular 3d object detection with depth-aware transformer,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 4012–4021

    2022

  2. [10]

    Attention is all you need,

    A. Vaswani, “Attention is all you need,” Advances in Neural Informa- tion Processing Systems , 2017

    2017

  3. [11]

    Monodetr: Depth-guided transformer for monocular 3d object detection,

    R. Zhang, H. Qiu, T. Wang, Z. Guo, Z. Cui, Y . Qiao, H. Li, and P. Gao, “Monodetr: Depth-guided transformer for monocular 3d object detection,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 9155–9166

    2023

  4. [12]

    Categorical depth distribution network for monocular 3d object detection,

    C. Reading, A. Harakeh, J. Chae, and S. L. Waslander, “Categorical depth distribution network for monocular 3d object detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 8555–8564

    2021

  5. [13]

    Dab-detr: Dynamic anchor boxes are better queries for detr,

    S. Liu, F. Li, H. Zhang, X. Yang, X. Qi, H. Su, J. Zhu, and L. Zhang, “Dab-detr: Dynamic anchor boxes are better queries for detr,” arXiv preprint arXiv:2201.12329, 2022

    2022 arXiv

  6. [14]

    Imagenet large scale visual recognition challenge,

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernsteinet al., “Imagenet large scale visual recognition challenge,” International journal of computer vision, vol. 115, pp. 211–252, 2015

    2015

  7. [15]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 770–778

    2016

  8. [16]

    Densely connected convolutional networks,

    G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, “Densely connected convolutional networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 4700–4708

    2017

  9. [17]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark et al., “Learning transferable visual models from natural language supervision,” in International conference on machine learning . PMLR, 2021, pp. 8748–8763. Fig. 7: Qu...

    2021

  10. [18]

    Emerging properties in self-supervised vision trans- formers,

    M. Caron, H. Touvron, I. Misra, H. J ´egou, J. Mairal, P. Bojanowski, and A. Joulin, “Emerging properties in self-supervised vision trans- formers,” in Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 9650–9660

    2021

  11. [19]

    Segment anything,

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y . Lo et al. , “Segment anything,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 4015–4026

    2023

  12. [20]

    Vision transformers for dense prediction,

    R. Ranftl, A. Bochkovskiy, and V . Koltun, “Vision transformers for dense prediction,” in Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 12 179–12 188

    2021

  13. [21]

    Depth anything v2,

    L. Yang, B. Kang, Z. Huang, Z. Zhao, X. Xu, J. Feng, and H. Zhao, “Depth anything v2,” arXiv preprint arXiv:2406.09414 , 2024

    2024 arXiv

  14. [22]

    Refinenet: Multi-path refinement networks for high-resolution semantic segmentation,

    G. Lin, A. Milan, C. Shen, and I. Reid, “Refinenet: Multi-path refinement networks for high-resolution semantic segmentation,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1925–1934

    2017

  15. [23]

    Monoc- ular relative depth perception with web stereo data supervision,

    K. Xian, C. Shen, Z. Cao, H. Lu, Y . Xiao, R. Li, and Z. Luo, “Monoc- ular relative depth perception with web stereo data supervision,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 311–320

    2018

  16. [24]

    Focal loss for dense object detection,

    T. Lin, “Focal loss for dense object detection,” arXiv preprint arXiv:1708.02002, 2017

    2017 arXiv

  17. [25]

    The hungarian method for the assignment problem,

    H. W. Kuhn, “The hungarian method for the assignment problem,” Naval research logistics quarterly , vol. 2, no. 1-2, pp. 83–97, 1955

    1955

  18. [26]

    3d object proposals for accurate object class detection,

    X. Chen, K. Kundu, Y . Zhu, A. G. Berneshawi, H. Ma, S. Fidler, and R. Urtasun, “3d object proposals for accurate object class detection,” Advances in neural information processing systems , vol. 28, 2015

    2015

  19. [27]

    Monocular 3d object detection for autonomous driving,

    X. Chen, K. Kundu, Z. Zhang, H. Ma, S. Fidler, and R. Urtasun, “Monocular 3d object detection for autonomous driving,” in Pro- ceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 2147–2156

    2016

  20. [28]

    Pointpillars: Fast encoders for object detection from point clouds,

    A. H. Lang, S. V ora, H. Caesar, L. Zhou, J. Yang, and O. Beijbom, “Pointpillars: Fast encoders for object detection from point clouds,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 12 697–12 705

    2019

  21. [29]

    Decoupled weight decay regularization,

    I. Loshchilov, “Decoupled weight decay regularization,” arXiv preprint arXiv:1711.05101, 2017

    2017 arXiv

  22. [30]

    Monocd: Monocular 3d object detection with complementary depths,

    L. Yan, P. Yan, S. Xiong, X. Xiang, and Y . Tan, “Monocd: Monocular 3d object detection with complementary depths,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 10 248–10 257

    2024

  23. [31]

    Exploring plain vision transformer backbones for object detection,

    Y . Li, H. Mao, R. Girshick, and K. He, “Exploring plain vision transformer backbones for object detection,” in European conference on computer vision . Springer, 2022, pp. 280–296

    2022

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.