Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

LMDepth: Lightweight Mamba-based Monocular Depth Estimation for Real-World Deployment

T0 review · 5 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 LMDepth, a lightweight Mamba-based monocular depth estimator, claims the best accuracy among lightweight methods on NYUDv2 and KITTI while using fewer parameters and lower FLOPs.

desk verdict The KITTI baseline table appears to contain NYUDv2 numbers copied with columns swapped, so the headline claim is currently unsupported, though the architecture and ablation are worth a second look. read the letter →

arxiv 2505.00980 v1 pith:OY345KSM submitted 2025-05-02 cs.CV

Jiahuan Long , Xin Zhou This is my paper · ORCID
classification cs.CV
keywords monoculardepthestimationMambastatespacemodelslightweightnetworkbinsINT8quantizationembeddeddeployment
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 establish that a state-space (Mamba) decoder can replace the heavy attention modules of Transformer-based depth estimators without sacrificing accuracy. Its model, LMDepth, combines a MobileNetV2 encoder with a modified pyramid spatial pooling head that outputs scene class, depth bins, and global features, and a decoder built from Depth Mamba Blocks. On NYUDv2 and KITTI, the authors report the best metrics among lightweight methods while using fewer parameters and GFLOPs (for example, δ1 = 0.854 and REL = 0.123 at 480×640 with 2.9M parameters and 2.77 GFLOPs). They further show that INT8 quantization shrinks the model to 2.63 MB with minimal accuracy loss. If correct, this gives resource-constrained robots and autonomous systems a depth-estimation backbone that is both accurate and cheap to run.

What carries the argument

The Depth Mamba Block (DMB) is the carrier of the argument: a decoder unit built on Vmamba modules that process image and depth tokens with a one-dimensional convolution followed by a selective state-space model, giving a global receptive field with cost linear in sequence length. Alongside it, the Modified Pyramid Spatial Pooling (MPSP) head pools at scales {1, 2, 3, 6} (or {1, 6} in the small variant) to produce scene classification, depth distribution bins, and global features. The mechanism's work is to let a lightweight network capture global context and discretize depth into bins, then reconstruct depth as probability-map-weighted bin values.

What would settle it

Re-run the compared lightweight baselines (FastDepth, GuideDepth, DaNet, TuMDE) on the same train/validation split and resolutions used for LMDepth on NYUDv2 and KITTI without changing anything else; if any baseline matches or exceeds LMDepth's δ1 or REL at equal or lower FLOPs, the central claim of a new efficiency frontier is falsified. A complementary check is to measure actual per-model latency on the same embedded board, since GFLOPs do not always translate to speed.

Watch

Extended reading notes

Core claim

The central claim is that LMDepth achieves the best performance across all evaluation metrics among lightweight monocular depth estimation methods on both NYUDv2 and KITTI, while requiring fewer parameters and lower GFLOPs. The architecture gets this from two components: a Modified Pyramid Spatial Pooling (MPSP) head that aggregates multi-scale context and predicts scene classification plus depth bins, and a decoder of Depth Mamba Blocks (DMBs) that fuses depth and visual features through linear-complexity state-space layers. The final depth is obtained by element-wise multiplication of a per-pixel probability map with the predicted bins. The authors also report that ONNX conversion plus INT8 quantization compresses the model to 2.63 MB and raises inference speed to 122 FPS on an embedded GPU platform, supporting deployment on resource-constrained devices.

Load-bearing premise

The reported advantage rests on the assumption that the baseline numbers cited from earlier papers were produced under protocols equivalent enough to LMDepth's — same data split, resolution, and preprocessing — so that cross-table comparisons are apples-to-apples.

Editorial extensions

If this is right

  • Mamba-style linear state-space decoders can replace quadratic-attention decoders in dense prediction: on NYUDv2, the Transformer decoder needs 2.8 GFLOPs versus 0.73 GFLOPs for the Mamba decoder, with similar or better accuracy.
  • A lightweight depth network can keep global context: the MPSP and DMB design reaches δ1 = 0.854 on NYUDv2 at 480×640 with only 2.9M parameters, a point on the accuracy-efficiency frontier.
  • The same network transfers across indoor and outdoor benchmarks: NYUDv2 δ1 = 0.854 and KITTI δ1 = 0.926 at high resolution, and zero-shot generalization on self-collected scenes beats the compared lightweight methods.
  • Practical edge deployment is feasible: ONNX plus INT8 quantization reduces the model to about 10% of its original size and reaches 122 FPS on an embedded platform, with δ2 dropping only from 0.86 to 0.84.
  • The auxiliary scene-classification branch provides a lightweight way to guide depth-bin generation, suggesting that semantic context and depth prediction can share a single compact head.

Reading between the lines

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

  • Editorial inference: Because the MPSP head outputs scene classification, the same backbone could naturally extend to joint depth estimation and scene understanding; the paper only uses classification as an auxiliary loss.
  • Editorial inference: The reported efficiency gains are measured in parameters and GFLOPs; actual latency on other edge devices, especially CPUs or lower-power NPUs, may not scale identically because Mamba's hardware-aware efficiency is partly GPU-oriented.
  • Editorial inference: The bin-based depth representation suggests a direct route to uncertainty estimation: the probability map over bins could be reinterpreted as a per-pixel depth distribution, yielding confidence intervals without extra supervision.
  • Editorial inference: A head-to-head re-run of baselines under identical splits, resolutions, and preprocessing would settle whether the reported margin is architectural or partly due to protocol differences; the paper relies on published numbers.
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

5 major / 5 minor

Summary. The paper proposes LMDepth, a lightweight Mamba-based monocular depth estimation network for embedded deployment. The architecture uses a MobileNetV2 encoder, a modified pyramid spatial pooling (MPSP) module that outputs scene classification and depth bins, and a decoder composed of Depth Mamba Blocks (DMBs). The authors report state-of-the-art accuracy-efficiency trade-offs on NYUDv2 and KITTI, with lower parameters and FLOPs than prior lightweight methods, and additionally present a deployment study on a Jetson AGX Xavier with INT8 quantization and zero-shot generalization on a self-collected indoor dataset. The central claim is that LMDepth achieves the best performance across all metrics among lightweight depth estimation methods on both benchmarks while remaining computationally lighter.

Significance. If the reported results are reliable, LMDepth would be a valuable contribution to efficient monocular depth estimation, extending Mamba-based architectures to dense prediction with a favorable accuracy-cost balance. The paper includes a practical deployment study, which is a strength relative to many architecture papers, and the proposed MPSP and DMB components are clearly described. However, the central comparative claims currently rest on baseline tables that contain at least one concrete internal inconsistency, and the evaluation protocol is not specified in sufficient detail to allow independent verification. The lack of released code or the self-collected evaluation set further limits reproducibility. The architectural idea and empirical direction are promising, but the evidence as printed does not yet fully support the headline 'best across all metrics' claim.

major comments (5)
  1. [Table II, Section IV.B] The low-resolution FastDepth row in the KITTI table reports δ1=0.778, δ2=0.942, δ3=0.987, REL=0.142, Sq-rel=0.591, RMSE=—, which is numerically identical to the low-resolution JointDepth row in Table I (NYUDv2) with the last two columns interchanged (JointDepth: δ1=0.778, δ2=0.942, δ3=0.987, REL=0.142, log10=—, RMS=0.591). This indicates that at least one KITTI baseline entry was copied or transposed from a different dataset/table rather than produced under the stated evaluation protocol. Since the paper's central claim of 'best performance across all evaluation metrics' on KITTI depends on the correctness of these baseline numbers, the authors should either correct Table II to match the original FastDepth results on KITTI or re-run the baseline under the same protocol and report the resulting numbers.
  2. [Section IV.A and Tables I-II] No evaluation protocol is specified for either dataset: the train/validation/test split (e.g., the standard Eigen split for KITTI or the official split for NYUDv2), the depth range or cap, the cropping scheme (e.g., Eigen crop, Garg crop, or center crop), and the preprocessing applied to RGB and depth images are all absent. Because no baseline is re-run in the authors' codebase and all baseline numbers are cited from previous papers, differences in these settings can easily account for the reported margins. The paper should state the exact protocol for both datasets and, ideally, verify at least one or two baselines (e.g., GuideDepth and DaNet) under the same protocol before claiming superiority across all metrics.
  3. [Section IV.C and Figure 5] The architecture variants (projection feature dimension, number of MPSP pooling scales, and the choice of LMDepth vs. LMDepth-S) are selected based on accuracy and FLOPs trade-offs that appear to be evaluated on the same benchmark that is later reported as the main result. The paper does not describe a separate validation split used during model selection. This makes the reported test numbers selection-aware rather than independent, inflating the apparent advantage. The authors should clarify which split was used for the ablation/selection curves and report the corresponding test-set results in a way that does not reuse the selection data.
  4. [Section V.B and Table V] There are contradictory deployment numbers. The text states that converting to ONNX 'reduces the model size from 26MB to 8.54MB', but Table V lists the PyTorch model size as 8.54 MB, with the ONNX size as 4.22 MB and the quantized size as 2.63 MB. Additionally, the text refers to a 'pruned LMDepth-T model' achieving 120 FPS, but no model named LMDepth-T is defined anywhere in the paper, and Table IV lists the same FPS for 'LMDepth (Ours)'. The deployment claims should be corrected and the naming clarified, since the claimed 4x compression and 2x speed-up do not match the numbers in Table V as printed.
  5. [Equation (5), Section III.E] The classification loss is written as Lcls(y, y_hat) = log(y_hat_k), with y_hat representing the predicted probabilities and y the ground truth. As a loss to be minimized, this should be the negative log-likelihood, i.e., -log(y_hat_k) for the true class k. If the implementation actually uses the negative log (as is standard for cross-entropy), the equation should be corrected; as written, minimizing Lcls would drive the predicted probability of the true class toward zero, which is inconsistent with the reported successful training. This is a load-bearing reproducibility issue.
minor comments (5)
  1. [Section IV.B, paragraph after Table I] The text says 'far lower than methods like GuideDepth (5.72 GFLOPs)', but Table I lists the high-resolution GuideDepth FLOPs as 10.47 G (and low-resolution as 2.63). Please correct this number to match the table.
  2. [Section V.A] The heading 'Data Acquisiotion Device' contains a typo; it should be 'Data Acquisition Device'.
  3. [Section V.B] The term 'pruned LMDepth-T' is introduced without any prior definition or explanation of pruning; if a pruned model is used for the FPS comparison, its construction should be described, or the name should be changed to match the model in Table IV.
  4. [Figure 2, Section I] The figure caption and text use inconsistent capitalization of 'MAMBA' (all caps) versus 'Mamba'; standardize to the conventional 'Mamba'.
  5. [Section IV.A, datasets] The dataset descriptions are very brief: the KITTI description does not mention the number of scenes or frames used, and the NYUDv2 description does not specify the split (e.g., 249 training scenes / 654 test images commonly used). Adding these details would improve reproducibility.

Circularity Check

2 steps flagged · score 2.0 of 10

The central accuracy claim is an external benchmark comparison and is not derived from the framework; the only circular steps are a self-referential discretization equation and the explicit selection of the reported model as the best variant on the same benchmark.

  1. self definitional [Section III-A, Eq. (2)]
    "A = exp(∆A), B = (∆A)−1(exp(∆A)− I)· ∆B. (2)"

    As printed, the discrete input matrix B is defined in terms of itself: the left-hand side B is the quantity being defined, while the right-hand side contains ∆B with the same symbol B. This is a self-referential definition rather than a closed-form ZOH discretization. It is a background preliminaries statement rather than a load-bearing part of the architecture's empirical evaluation, but as written it is circular.

  2. fitted input called prediction [Section IV-C, Figure 5 and Table I]
    "We select the model with the highest accuracy as LMDepth, and select the model with a point of maximum gradient as LMDepth-S."

    The reported flagship model LMDepth is selected by construction as the highest-accuracy point from the architecture sweep evaluated on the same depth-estimation benchmark that is later reported as the main result. Thus LMDepth's superiority over its own architecture variants is definitional rather than predictive. This does not by itself force the headline comparison against external baselines, which is an independent empirical comparison, but it is a genuine selection-on-the-reported-benchmark circularity.

full rationale

The paper's central claim—that LMDepth achieves better accuracy with fewer parameters and lower GFLOPs than prior lightweight methods on NYUDv2 and KITTI—is an empirical comparison against externally published baselines. That claim is not derived from the Mamba equations or from the MPSP/DMB design; it stands or falls on the correctness and comparability of the benchmark tables. No uniqueness theorem, self-citation, or parameter-fit chain forces the result. The two circular elements are minor: Eq. (2) defines B in terms of B as printed, and the authors explicitly choose LMDepth as the highest-accuracy point from a sweep on the same benchmark before reporting its numbers. The latter inflates the significance of the internal ablation but does not by construction determine the external comparison. The KITTI table contains a likely transcription error in the FastDepth row, but that is a data-integrity issue, not circularity. Overall, the core derivation is self-contained and empirically tested, so the circularity score is low.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claim rests on several unverified domain assumptions: the benefit of VMamba over CNN/Transformer decoders, the sufficiency of MobileNetV2 features, the usefulness of scene classification for depth bin generation, and the direct comparability of external baseline numbers. Several key hyperparameters (loss weights, projection dimensions, pooling scales, bin count) are manually chosen or selected on the evaluation benchmark itself, and the bin count is not even disclosed.

free parameters (6)
  • alpha (depth regression loss weight) = 10
    Fixed for all experiments following AdaBins [9]; a manual hyperparameter.
  • lambda (scale-invariant loss term) = 0.85
    Taken from AdaBins [9]; set by hand.
  • beta (classification loss weight) = 0.1
    Set by hand in Implementation Details.
  • Decoder projection dimension = 24 (LMDepth), 16 (LMDepth-S)
    Chosen via ablation in Section IV.C using accuracy/FLOPs trade-off; a fitted architectural hyperparameter on the evaluation benchmark.
  • MPSP pooling scales = {1,2,3,6} (LMDepth), {1,6} (LMDepth-S)
    Ablated in Section IV.C and selected on the same benchmark, effectively a hand-tuned architectural choice.
  • Number of depth bins = Not specified
    The adaptive-bins head's bin count is a free design parameter but is never stated, so it cannot be audited.
assumptions (4)
  • domain assumption The VMamba module [16] provides linear-complexity global feature modeling and improves over CNN and Transformer decoders for depth.
    Invoked in Section III.C as the basis for DMB; the paper's own ablation (Table III) is the only supporting evidence, not an externally verified fact.
  • domain assumption MobileNetV2 features are sufficient as the sole encoder for both indoor and outdoor monocular depth estimation.
    Used throughout; no comparison against other encoders is given.
  • domain assumption The 25 NYUDv2 folder-name classes provide meaningful auxiliary supervision for depth bin generation.
    Section IV.A states the 25 classes are used for scene classification; validity for depth is assumed.
  • domain assumption Numbers reported for baseline methods from prior papers are directly comparable to this paper's results despite different training setups, resolutions, and codebases.
    All comparisons in Tables I, II, and IV rely on this; no baselines are re-run.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LMDepth: Lightweight Mamba-based Monocular Depth Estimation for Real-World Deployment." pith.science (2026). https://pith.science/paper/OY345KSM

@misc{pith2026250500980,
  author       = {Pith},
  title        = {Pith review of: LMDepth: Lightweight Mamba-based Monocular Depth Estimation for Real-World Deployment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OY345KSM}},
  note         = {Machine review of arXiv:2505.00980}
}
read the original abstract

Monocular depth estimation provides an additional depth dimension to RGB images, making it widely applicable in various fields such as virtual reality, autonomous driving and robotic navigation. However, existing depth estimation algorithms often struggle to effectively balance performance and computational efficiency, which poses challenges for deployment on resource-constrained devices. To address this, we propose LMDepth, a lightweight Mamba-based monocular depth estimation network, designed to reconstruct high-precision depth information while maintaining low computational overhead. Specifically, we propose a modified pyramid spatial pooling module that serves as a multi-scale feature aggregator and context extractor, ensuring global spatial information for accurate depth estimation. Moreover, we integrate multiple depth Mamba blocks into the decoder. Designed with linear computations, the Mamba Blocks enable LMDepth to efficiently decode depth information from global features, providing a lightweight alternative to Transformer-based architectures that depend on complex attention mechanisms. Extensive experiments on the NYUDv2 and KITTI datasets demonstrate the effectiveness of our proposed LMDepth. Compared to previous lightweight depth estimation methods, LMDepth achieves higher performance with fewer parameters and lower computational complexity (measured by GFLOPs). We further deploy LMDepth on an embedded platform with INT8 quantization, validating its practicality for real-world edge applications.

Figures

Figures reproduced from arXiv: 2505.00980 by the authors.

Figure 1
Figure 1. Complextity and performance comparisons of representative lightweight depth estimation methods. The size of ◦ and ∆ represents GFlops, with larger sizes indicating higher GFlops. Compared with other methods, our method has higher accuracy while keeping fewer parameters and lower GFlops across different image resolutions. improving accuracy while often overlooking the challenge of deploying on resource-constrained de… view at source ↗
Figure 2
Figure 2. Comparison of computational processes of tokens among CNN, Transformer, and MAMBA. It highlights why our Mamba-based depth estimation method is both efficient and powerful. Compared to CNN, MAMBA achieves an improved receptive field. In contrast to Transformer, MAMBA reduces computational complexity from quadratic to linear. II. RELATED WORK A. Lightweight Monocular Depth Estimation Early monocular depth estimation … view at source ↗
Figure 3
Figure 3. The overall network architecture of our LMdepth. It consists of an image encoder, an MPSP head, and a decoder. The MPSP head outputs predicted scene classification and depth bins, while the decoder, composed of four custom-designed DMBs, predicts the depth distribution probabilities. [30], [31] and image dehazing [32], demonstrating strong adaptability and efficiency. Given Mamba’s balance between performance and ef… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The overview of VMamba Module. A one-dimensional convo￾lution operation is added to the original Mamba to enhance the extraction of local attention. B. Overall Architecture [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Varing projection features on 2 and 4 MPSP layers. Red lines show the trend of δ1 accuracy, and Blue lines show trend of FLOPs. LMDepth is selected for the highest accuracy, and LMDepth-S is selected for the equilibrium point between efficiency and performance. C. Abla…
Figure 6
Figure 6. Figure 6: Qualitative comparison of various lightweight depth estimation methods on the indoor and outdoor scenarios. Our predictions are more refined and closer to ground truth. It demonstrates that our method achieves better depth estimation of distant objects and finer detail…
Figure 7
Figure 7. Figure 7: Embedded deployment platform and data acquisition pipeline. (a) The lightweight deployment platform is based on the NVIDIA Jetson AGX Xavier, equipped with a Volta GPU featuring 512 CUDA cores and 64 Tensor cores. (b) The mobile data acquisition device comprises a LiDA…
Figure 8
Figure 8. Figure 8: Qualitative comparison of various lightweight depth estimation methods on real-world scenarios. Our predictions are more refined and closer to ground truth. It demonstrates that our method achieves better depth estimation of distant objects and finer details. achieves …

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. DepthART: Scaling Foundation Monocular Depth to Tiny Models

    cs.CV 2026-07 conditional novelty 6.0 of 10

    DepthART transfers foundation-style monocular depth to 6-33M-parameter models via bias-resistant data sampling and frozen-encoder camera-conditioned metric fine-tuning, reaching near-foundation zero-shot accuracy at r...

Reference graph

Works this paper leans on

41 extracted references · 15 canonical work pages · cited by 1 Pith paper

  1. [1]

    Splat-slam: Globally optimized rgb-only slam with 3d gaussians,

    E. Sandstr ¨om, K. Tateno, M. Oechsle, M. Niemeyer, L. Van Gool, M. R. Oswald, and F. Tombari, “Splat-slam: Globally optimized rgb-only slam with 3d gaussians,” arXiv preprint arXiv:2405.16544 , 2024

    arXiv 2024

  2. [2]

    Pseudo rgb- d for self-improving monocular slam and depth prediction,

    L. Tiwari, P. Ji, Q. H. Tran, B. Zhuang, and M. Chandraker, “Pseudo rgb- d for self-improving monocular slam and depth prediction,” in European Conference on Computer Vision (ECCV) , 2020

    work page 2020

  3. [3]

    Physical 3d adversarial attacks against monocular depth estimation in autonomous driving,

    J. Zheng, C. Lin, J. Sun, Z. Zhao, Q. Li, and C. Shen, “Physical 3d adversarial attacks against monocular depth estimation in autonomous driving,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 24 452–24 461

    work page 2024

  4. [4]

    Mgnet: Monocular geo- metric scene understanding for autonomous driving,

    M. Sch ¨on, M. Buchholz, and K. Dietmayer, “Mgnet: Monocular geo- metric scene understanding for autonomous driving,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 15 804–15 815

    2021

  5. [5]

    J-mod 2: Joint monocular obstacle detection and depth estimation,

    M. Mancini, G. Costante, P. Valigi, and T. A. Ciarfuglia, “J-mod 2: Joint monocular obstacle detection and depth estimation,” IEEE Robotics and Automation Letters, vol. 3, no. 3, pp. 1490–1497, 2018

    work page 2018

  6. [6]

    Deeper depth prediction with fully convolutional residual networks,

    I. Laina, C. Rupprecht, V . Belagiannis, F. Tombari, and N. Navab, “Deeper depth prediction with fully convolutional residual networks,” in 2016 Fourth international conference on 3D vision (3DV) . IEEE, 2016, pp. 239–248

    2016

  7. [7]

    Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer,

    K. Lasinger, R. Ranftl, K. Schindler, and V . Koltun, “Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer,” arXiv preprint arXiv:1907.01341 , 2019

    arXiv 1907

  8. [8]

    Monocular depth estimation using laplacian pyramid-based depth residuals,

    M. Song, S. Lim, and W. Kim, “Monocular depth estimation using laplacian pyramid-based depth residuals,” IEEE transactions on circuits and systems for video technology , vol. 31, no. 11, pp. 4381–4393, 2021

    2021

Show all 41 references
  1. [9]

    Adabins: Depth estimation using adaptive bins,

    S. F. Bhat, I. Alhashim, and P. Wonka, “Adabins: Depth estimation using adaptive bins,” in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2021

    2021

  2. [10]

    Neural window fully- connected crfs for monocular depth estimation,

    W. Yuan, X. Gu, Z. Dai, S. Zhu, and P. Tan, “Neural window fully- connected crfs for monocular depth estimation,” pp. 3916–3925, 2022

    2022

  3. [11]

    Iebins: Itera- tive elastic bins for monocular depth estimation,

    S. Shao, Z. Pei, X. Wu, Z. Liu, W. Chen, and Z. Li, “Iebins: Itera- tive elastic bins for monocular depth estimation,” Advances in Neural Information Processing Systems , vol. 36, 2024

    2024

  4. [12]

    Fastdepth: Fast monocular depth estimation on embedded systems,

    D. Wofk, F. Ma, T.-J. Yang, S. Karaman, and V . Sze, “Fastdepth: Fast monocular depth estimation on embedded systems,” in 2019 Interna- tional Conference on Robotics and Automation (ICRA) . IEEE, 2019, pp. 6101–6108

    2019

  5. [13]

    Efficient monocular depth estimation for edge devices in internet of things,

    X. Tu, C. Xu, S. Liu, R. Li, G. Xie, J. Huang, and L. T. Yang, “Efficient monocular depth estimation for edge devices in internet of things,” IEEE Transactions on Industrial Informatics , vol. 17, no. 4, pp. 2821–2832, 2020

    2020

  6. [14]

    Lightweight monocular depth estimation through guided decoding,

    M. Rudolph, Y . Dawoud, R. G¨uldenring, L. Nalpantidis, and V . Belagian- nis, “Lightweight monocular depth estimation through guided decoding,” in 2022 International Conference on Robotics and Automation (ICRA) . IEEE, 2022, pp. 2344–2350

    2022

  7. [15]

    Vision mamba: Efficient visual representation learning with bidirectional state space model,

    L. Zhu, B. Liao, Q. Zhang, X. Wang, W. Liu, and X. Wang, “Vision mamba: Efficient visual representation learning with bidirectional state space model,” arXiv preprint arXiv:2401.09417 , 2024

    2024 arXiv

  8. [16]

    Mambavision: A hybrid mamba- transformer vision backbone,

    A. Hatamizadeh and J. Kautz, “Mambavision: A hybrid mamba- transformer vision backbone,” arXiv preprint arXiv:2407.08083 , 2024

    2024 arXiv

  9. [17]

    Remam- ber: Referring image segmentation with mamba twister,

    Y . Yang, C. Ma, J. Yao, Z. Zhong, Y . Zhang, and Y . Wang, “Remam- ber: Referring image segmentation with mamba twister,” in European Conference on Computer Vision . Springer, 2025, pp. 108–126

    2025

  10. [18]

    U-mamba: Enhancing long-range dependency for biomedical image segmentation,

    J. Ma, F. Li, and B. Wang, “U-mamba: Enhancing long-range dependency for biomedical image segmentation,” arXiv preprint arXiv:2401.04722, 2024

    2024 arXiv

  11. [19]

    Make3d: Learning 3d scene structure from a single still image,

    A. Saxena, M. Sun, and A. Y . Ng, “Make3d: Learning 3d scene structure from a single still image,” IEEE transactions on pattern analysis and machine intelligence, vol. 31, no. 5, pp. 824–840, 2008

    2008

  12. [20]

    Discrete-continuous depth estimation from a single image,

    M. Liu, M. Salzmann, and X. He, “Discrete-continuous depth estimation from a single image,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2014, pp. 716–723

    2014

  13. [21]

    Depth map prediction from a single image using a multi-scale deep network,

    D. Eigen, C. Puhrsch, and R. Fergus, “Depth map prediction from a single image using a multi-scale deep network,” Advances in neural information processing systems , vol. 27, 2014

    2014

  14. [22]

    Predicting depth, surface normals and se- mantic labels with a common multi-scale convolutional architecture,

    D. Eigen and R. Fergus, “Predicting depth, surface normals and se- mantic labels with a common multi-scale convolutional architecture,” in Proceedings of the IEEE international conference on computer vision , 2015, pp. 2650–2658

    2015

  15. [23]

    Towards robust monocular depth estimation: Mixing datasets for zero-shot cross- dataset transfer,

    R. Ranftl, K. Lasinger, D. Hafner, K. Schindler, and V . Koltun, “Towards robust monocular depth estimation: Mixing datasets for zero-shot cross- dataset transfer,” IEEE transactions on pattern analysis and machine intelligence, vol. 44, no. 3, pp. 1623–1637, 2020

    2020

  16. [24]

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

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An image is worth 16x16 words: Transformers for image recognition at scale,” CoRR, vol. abs/2010.11929, 2020. [Onlin...

    2010 arXiv

  17. [25]

    Combining recurrent, convolutional, and continuous-time models with linear state space layers,

    A. Gu, I. Johnson, K. Goel, K. Saab, T. Dao, A. Rudra, and C. R ´e, “Combining recurrent, convolutional, and continuous-time models with linear state space layers,” Advances in neural information processing systems, vol. 34, pp. 572–585, 2021

    2021

  18. [26]

    Efficiently modeling long sequences with structured state spaces,

    A. Gu, K. Goel, and C. R ´e, “Efficiently modeling long sequences with structured state spaces,” CoRR, vol. abs/2111.00396, 2021. [Online]. Available: https://arxiv.org/abs/2111.00396

    2021 arXiv

  19. [27]

    Simplified state space layers for sequence modeling,

    J. T. H. Smith, A. Warrington, and S. W. Linderman, “Simplified state space layers for sequence modeling,” 2023. [Online]. Available: https://arxiv.org/abs/2208.04933

    2023 arXiv

  20. [28]

    Hungry hungry hippos: Towards language modeling with state space models,

    D. Y . Fu, T. Dao, K. K. Saab, A. W. Thomas, A. Rudra, and C. R ´e, “Hungry hungry hippos: Towards language modeling with state space models,” arXiv preprint arXiv:2212.14052 , 2022

    2022 arXiv

  21. [29]

    Mamba: Linear-time sequence modeling with selective state spaces,

    A. Gu and T. Dao, “Mamba: Linear-time sequence modeling with selective state spaces,” arXiv preprint arXiv:2312.00752 , 2023

    2023 arXiv

  22. [30]

    Segmamba: Long-range sequential modeling mamba for 3d medical image segmentation,

    Z. Xing, T. Ye, Y . Yang, G. Liu, and L. Zhu, “Segmamba: Long-range sequential modeling mamba for 3d medical image segmentation,” in International Conference on Medical Image Computing and Computer- Assisted Intervention. Springer, 2024, pp. 578–588

    2024

  23. [31]

    Vm-unet: Vision mamba unet for medical image segmentation,

    J. Ruan and S. Xiang, “Vm-unet: Vision mamba unet for medical image segmentation,” arXiv preprint arXiv:2402.02491 , 2024

    2024 arXiv

  24. [32]

    U-shaped vision mamba for single image dehazing,

    Z. Zheng and C. Wu, “U-shaped vision mamba for single image dehazing,” arXiv preprint arXiv:2402.04139 , 2024

    2024 arXiv

  25. [33]

    Inverted residuals and linear bottlenecks: Mobile networks for classification, detection and segmentation,

    M. Sandler, A. G. Howard, M. Zhu, A. Zhmoginov, and L. Chen, “Inverted residuals and linear bottlenecks: Mobile networks for classification, detection and segmentation,” vol. abs/1801.04381, 2018. [Online]. Available: http://arxiv.org/abs/1801.04381

    2018 arXiv

  26. [34]

    Pyramid scene parsing network,

    H. Zhao, J. Shi, X. Qi, X. Wang, and J. Jia, “Pyramid scene parsing network,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 2881–2890

    2017

  27. [35]

    Real-time joint semantic segmentation and depth estimation using asymmetric annotations,

    V . Nekrasov, T. Dharmasiri, A. Spek, T. Drummond, C. Shen, and I. Reid, “Real-time joint semantic segmentation and depth estimation using asymmetric annotations,” in 2019 International Conference on Robotics and Automation (ICRA) . IEEE, 2019, pp. 7101–7107

    2019

  28. [36]

    Enforcing geometric constraints of virtual normal for depth prediction,

    W. Yin, Y . Liu, C. Shen, and Y . Yan, “Enforcing geometric constraints of virtual normal for depth prediction,” in Proceedings of the IEEE/CVF international conference on computer vision , 2019, pp. 5684–5693

    2019

  29. [37]

    Monocular depth distribution alignment with low computation,

    F. Sheng, F. Xue, Y . Chang, W. Liang, and A. Ming, “Monocular depth distribution alignment with low computation,” in 2022 International Conference on Robotics and Automation (ICRA) . IEEE, 2022, pp. 6548–6555

    2022

  30. [38]

    Pytorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, et al. , “Pytorch: An imperative style, high-performance deep learning library,” Advances in neural information processing systems , vol. 32, 2019

    2019

  31. [39]

    Adam: A method for stochastic optimization,

    P. K. Diederik, “Adam: A method for stochastic optimization,” (No Title), 2014

    2014

  32. [40]

    Indoor segmentation and support inference from rgbd images,

    N. Silberman, D. Hoiem, P. Kohli, and R. Fergus, “Indoor segmentation and support inference from rgbd images,” in Computer Vision–ECCV 2012: 12th European Conference on Computer Vision, Florence, Italy, October 7-13, 2012, Proceedings, Part V 12 . Springer, 2012, pp. 746– 760

    2012

  33. [41]

    Vision meets robotics: The kitti dataset,

    A. Geiger, P. Lenz, C. Stiller, and R. Urtasun, “Vision meets robotics: The kitti dataset,” The International Journal of Robotics Research , vol. 32, no. 11, pp. 1231–1237, 2013

    2013

Pith tools

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