Pith. sign in

REVIEW 4 major objections 5 minor 45 references

TSceneJAL: Joint Active Learning of Traffic Scenes for 3D Object Detection

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read TSceneJAL proposes that a three-stage sampler — category entropy, graph-based scene similarity, and mixture-density-network uncertainty — selects traffic scenes that train better 3D object detectors than existing active learning methods…

desk verdict A workmanlike empirical AL paper: known ingredients, new integration, consistent moderate gains on four datasets; theory is weak and early-round pseudo-label noise is under-examined, but the empirical core probably survives. read the letter →

arxiv 2412.18870 v3 pith:O7EQWNH2 submitted 2024-12-25 cs.CV

classification cs.CV
keywords activelearning3Dobjectdetectionautonomousdrivingcategoryentropyscenesimilaritymixturedensitynetworkuncertaintyestimationpointclouds
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

TSceneJAL aims to solve a practical problem: under a fixed annotation budget, which unlabeled traffic scenes should be labeled to train the best 3D object detector. The paper's answer is a three-stage sampler that filters scenes in sequence — first those with the most balanced object-category mix, then the most diverse among those by graph-based scene similarity, then the most complex by a mixture-density-network uncertainty score. The paper claims this joint policy beats Random, Confidence, MC Dropout, Coreset, Badge, and Crb on KITTI, Lyft, nuScenes, and SUScape, with average mAP3D gains over Random of about 2.45%, 2.13%, 1.24%, and 1.25%, respectively, and gains over the strongest prior method, Crb, between roughly 0.9 and 1.6 points. The practical stakes are that labeling 3D driving data is expensive, and the paper also reports that on KITTI its selected sets contain roughly half the objects of Random's sets at comparable or better accuracy, and that sampling 60–80% of the data can match fully supervised training.

What carries the argument

The mechanism is the three-stage cascade itself. Stage one ranks unlabeled scenes by Eq. (12), a class-proportion entropy of pseudo-labels thresholded at confidence 0.3, and keeps the top $K_1N_r$; stage two builds a directed scene graph with one node per detected object plus the ego vehicle, edge weights $1/\text{distance}$, computes scene similarity through the marginalized kernel of Eq. (17), and applies farthest-point sampling (Algorithm 1) to keep $K_2N_r$ diverse scenes; stage three scores each remaining scene with Eq. (23), the mean over boxes and seven box attributes of aleatoric plus epistemic variance from the MDN regression head, and keeps the $N_r$ most uncertain. Algorithm 2 stitches these stages together and retrains the PointPillars predictor each round.

What would settle it

Run the same three-stage sampler twice from a very small initial labeled pool — once scoring scenes from the detector's pseudo-labels and once scoring the same scenes from oracle ground-truth labels — and compare the mAP3D of the final trained detectors; if the pseudo-label version loses most or all of its advantage over Random in early rounds, the central claim that the three metrics are robust selection signals would be falsified.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that optimal active learning for 3D detection should be treated as a three-part divergence-minimization problem — balance, diversity, and complexity — and that a stage-by-stage greedy cascade is the right way to minimize all three at once. Each scene is scored three ways: category entropy of the predicted class distribution filters for balance; a directed graph whose nodes are detected objects and whose edges encode inverse 3D distance, compared with the marginalized kernel and sampled by farthest-point selection, filters for diversity; and a mixture density network attached to the regression head outputs per-attribute aleatoric and epistemic variances whose sum scores complexity. Running these metrics in the order entropy → similarity → uncertainty produces the reported results: consistent mAP3D/mAPBEV improvements over all baselines on all four datasets, with the largest redundancy-related gains on the sequential Lyft data and up to 12% improvement reported overall.

Load-bearing premise

The pipeline assumes that the current detector's pseudo-labels — class confidences, box regressions, and MDN variances — are reliable enough that the three metrics rank scenes correctly, an assumption that is most strained in the first active-learning rounds when the labeled pool is tiny and false positives can, as the paper's Fig. 9 shows, turn a one-car scene into a high-entropy scene.

Editorial extensions

If this is right

  • If the central claim holds, a fixed annotation budget buys more detection accuracy: the paper's TSceneJAL beats Random by 2.45 mAP3D on KITTI, 2.13 on Lyft, 1.24 on nuScenes, and 1.25 on SUScape at the end of the active learning process.
  • The reported KITTI object-count statistics imply annotation cost is roughly halved for the same model quality, because TSceneJAL's selected scenes contain about half as many objects as Random's and still yield higher mAP.
  • Sequential, redundant data benefits most from the joint policy: the Lyft dataset, collected as continuous drives, shows the largest similarity-stage gains (+1.61 to +2.01 mAP3D over the no-similarity variant), suggesting redundancy removal is where the biggest wins lie.
  • The stage order matters: entropy-first beats uncertainty-first by a clear margin in Table VIII, so practitioners should prioritize class balance before complexity sampling.
  • On KITTI, matching or exceeding fully supervised training requires only 60–80% of the data, implying roughly 20–40% of many existing AD datasets may be redundant for training.

Reading between the lines

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

  • A testable extension is to make the confidence threshold of Eq. (14) adaptive across active-learning rounds; the paper's own sweep shows a 0.9 threshold degrades to near-random, so a noise-aware threshold could extend the gains to very early rounds where pseudo-labels are least reliable.
  • If pseudo-label noise is the binding constraint, coupling the three-stage sampler with a pseudo-label refinement step — for example, ensembling or temporal aggregation of predictions before scoring — could preserve more of the advantage in low-budget regimes, a direction the paper leaves open.
  • Because the scene-graph similarity uses only 3D positions and class names, the same diversity stage should transfer to other geometry-heavy perception tasks, such as indoor 3D object detection or LiDAR segmentation, where redundant frames also inflate labeling costs.
  • The KITTI redundancy result (60–80% of data matching fully supervised accuracy) suggests a standalone benchmark: measure how much pure near-duplicate removal from raw sensor streams can cut annotation budgets before any active-learning loop is added.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes TSceneJAL, an active learning (AL) framework for 3D object detection in autonomous driving. The method combines three sampling criteria in a fixed three-stage pipeline: category entropy to encourage class balance, graph-based scene similarity with farthest-point sampling to reduce redundancy, and mixture density network (MDN) uncertainty to select complex scenes. The AL predictor is a PointPillars model with the regression head modified to an MDN, so that aleatoric and epistemic uncertainties are obtained in one forward pass. The method is evaluated on KITTI, Lyft, nuScenes, and SUScape against Random, Confidence, MC Dropout, Coreset, Badge, and Crb. Table I reports that TSceneJAL achieves the best final mAP3D/mAPBEV on most settings, with gains over Random of roughly 1.2–2.5 mAP points depending on dataset and difficulty level. Ablations study each metric, the confidence threshold, stage order, initial data quantity, and selection complexity.

Significance. If the empirical claims hold, the contribution is a practical, multi-criteria AL pipeline for LiDAR-based 3D detection, with the useful engineering property that uncertainty estimation is obtained in a single forward pass. The paper's strengths are its breadth of datasets, the inclusion of a released code repository, and ablations that separately test each of the three metrics, the stage order, and the confidence threshold. The main limitations are that the headline comparison rests on single runs without error bars or significance tests, that the proposed mechanism is not verified in the low-label regime on which AL depends, and that the theoretical derivation in Section III is not sound as written. Overall, the empirical direction is promising and the issues are addressable, but the central claim needs additional support before publication.

major comments (4)
  1. [Section VII-B, Table I, and Table VI] The central empirical claim that TSceneJAL outperforms all baselines is not statistically supported. Table I reports only single-run mAP values, while the gains over Random are small (e.g., roughly 1.2 mAP3D points on nuScenes). The only repeated-run results in the paper, Table VI, show standard deviations of 0.65–1.47 mAP3D for a closely related selection pipeline, which is the same magnitude as several of the headline gains. Please report results over multiple seeds with means and error bars, and ideally paired significance tests, for at least the KITTI and nuScenes comparisons.
  2. [Section V (Eqs. 12–14), Section VII-C1, Figures 5 and 9] Pseudo-label noise in early AL rounds is acknowledged but never quantified. All three metrics are computed from the current predictor's pseudo-labels: category entropy uses predicted class counts, scene graphs use predicted classes and boxes, and MDN uncertainty comes from the same regression head. In the first KITTI round the predictor is trained on only 200 scenes, and Section VII-C1 states that the AL predictor is 'hardly capable of detecting objects in the initial learning process.' The confidence threshold in Eq. (14) and its ablation in Table X address one symptom, but the paper does not measure how pseudo-label noise degrades the entropy, similarity, or uncertainty rankings. Figure 5 shows that TSceneJAL is not better in the first two KITTI rounds and improves later, which is consistent with the metrics becoming reliable only after the predictor improves. Please add a quantitative analysis of ranking stability or pseudo-label noise sensitivity, or temper the mechanistic interpretation of the gains.
  3. [Section III, Eqs. (3), (8)–(10)] The theoretical derivation is not sound as written. Eq. (8) states D*_S3 = arg max H(DS3) = arg max [H(DS3|ω) + H(ω) − H(ω|DS3)], and Eq. (10) concludes D*_S3 ≈ arg min H(ω|DS3) from the assertion H(ω|DS3) ≪ H(DS3|ω). If the conditional entropy is negligible, dropping it from a maximization does not yield a minimization of that term; the conclusion would require the other two terms to be constant, which is neither stated nor established. In addition, Eq. (3) introduces a decomposition of the distribution discrepancy into three KL terms with assumed target distributions (uniform, Gaussian, uniform) and no derivation; this appears to be a post hoc mapping onto the three proposed metrics. Because the empirical comparison is independent of this derivation, the issue is fixable by rewriting Section III as a design rationale rather than a proof.
  4. [Abstract, Section VII-B, and Section VIII] The reported improvement ranges are inconsistent with Table I. The abstract promises improvements 'up to 12%,' and the conclusion lists average improvements of 2.5%–6.6%, 4.4%–9.0%, 3.0%–10.9%, and 0.5%–6.2% for the four datasets. Direct comparisons to Random in Table I are about 2.2–2.6 points for KITTI, 1.7–2.5 for Lyft, 1.0–1.5 for nuScenes, and 1.1–1.5 for SUScape; no value is near 9–12%. Please correct the abstract and conclusion, or specify exactly which baseline and metric produce the larger ranges.
minor comments (5)
  1. [Section V-A, Eq. (12)] The entropy term log(p_c + ζ) assigns a large positive entropy to a scene with no detections (p_c = 0 for all c), since H = −C log ζ. Please state how empty scenes are handled in the category entropy computation.
  2. [Section V-B, Eq. (15)–(17)] The kernel K(v, v') = ξ(v = v')/2 assigns similarity 1/2 between identical nodes rather than 1; please clarify whether this normalization is intentional, and correct the phrase 'marginalize kernel' to 'marginalized kernel' throughout the paper.
  3. [Section V-C, Eq. (22)] For the anchor-based residual encoding used by PointPillars, the propagation of variance for width, length, and height should include the anchor dimensions (or the actual box dimensions), not only the residual means squared. Please clarify the notation and verify the formula.
  4. [Section VII-C2, Table VII] The caption states that numerical values are presented in 10^-2 order, but the table lists values such as 0.663 and 1.732; please make the stated scale consistent with the displayed numbers.
  5. [General] The paper contains several typographical errors, including 'sence sampling' in the index terms, 'correponding' in the Fig. 4 caption, and 'pesudo-labels' in the introduction; a careful proofreading pass is recommended.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the empirical comparison is self-contained and the theoretical section, while post hoc, does not force the reported gains.

full rationale

The central claim is empirical: TSceneJAL selects scenes that improve 3D detection on four datasets relative to standard baselines, and this comparison is measured on oracle-labeled validation sets independently of the selection rule. None of the three metrics (category entropy, graph-kernel similarity, MDN uncertainty) is fitted to the reported mAP numbers, and the final stage-3 selection is not obtained by inverting the performance table. The theoretical derivation in Sec. III is a post-hoc decomposition: Eq. (3) defines the discrepancy as exactly the three KL terms (balance, diversity, complexity) that the proposed metrics then operationalize, so it is a motivating framework rather than an independent first-principles proof. That does not make the detection results circular, because the mapping from Eq. (3) to the implemented algorithms involves non-trivial choices (graph kernels, farthest sampling, MDN estimation) and the empirical gains are not entailed by the KL definitions. The paper's own admission in Sec. VII-C1 that the AL predictor is 'hardly capable of detecting objects in the initial learning process' (Fig. 9) is a genuine robustness limitation about pseudo-label noise in early rounds, but it is not a circularity: the metrics are computed from pseudo-labels, yet the reported performance is measured on labeled validation data. Self-citations (SUScape dataset [3], SUSTech Points annotation tool [14]) are benchmarks and tools, not load-bearing arguments. No load-bearing claim reduces by construction to a fitted parameter or to an author-supplied uniqueness theorem.

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

The central claim rests on several hand-set hyperparameters (tau, eta, K1, K2, sigma, K) and on unproved distributional assumptions in the theoretical section. No new physical entities are introduced; the mirror node in the graph representation is a bookkeeping construct, not a physical entity.

free parameters (5)
  • category confidence threshold tau = 0.3
    Used in Eq. (14) to filter predicted boxes before computing category entropy; tuned on KITTI in Table X.
  • uncertainty mixing ratio eta = 0.5
    Weight for epistemic uncertainty in Eq. (23); fixed without a reported sweep.
  • stage selection ratios K1, K2 = K1=3, K2=2.5
    Set in Sec. VII-A to produce 3N_r and 2.5N_r candidates in stages 1 and 2; no ablation over these ratios.
  • graph edge kernel bandwidth sigma = not specified
    Appears in the edge kernel in Eq. (16) as exp(-||e-e'||/2 sigma^2), but no value or estimation procedure is given.
  • number of MDN mixture components K = 3
    Chosen in Sec. VII-A as a tradeoff between training efficiency and detection performance; no ablation shown.
assumptions (5)
  • ad hoc to paper The distribution discrepancy in Eq. (1) decomposes into the sum of three KL divergences over category, graph similarity, and uncertainty distributions, with P_YT uniform, P_GT Gaussian, P_DT uniform (Eq. 3).
    This decomposition is asserted without proof and directly determines the three metrics; it is load-bearing for the theoretical motivation.
  • domain assumption The target category distribution P_YT of the test set is uniform.
    Used to reduce the first KL term to entropy maximization in Eq. (5). AD test sets are class-imbalanced, so this assumption is likely false.
  • ad hoc to paper In Eq. (8), H(omega|D_S3) is much smaller than H(D_S3|omega), allowing uncertainty selection to be derived from the entropy decomposition.
    The paper states this to justify focusing on the last term, but the inequality actually points toward maximizing H(D|omega), not minimizing H(omega|D); the transition is mathematically unsupported.
  • domain assumption Pseudo-labels from the current AL predictor are sufficiently reliable for category entropy and scene graph construction.
    All three metrics use predicted boxes and classes from a model trained on a small labeled pool; only a confidence threshold (Eq. 14) is used to mitigate errors, with no systematic validation.
  • domain assumption The marginalized kernel random walk and the chosen edge kernel (Eq. 16) provide a meaningful similarity measure between traffic scenes.
    The kernel is taken from Kashima et al. [40]; the paper does not validate that the similarity ranking correlates with downstream detection value, and the bandwidth sigma is unspecified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TSceneJAL: Joint Active Learning of Traffic Scenes for 3D Object Detection." pith.science (2026). https://pith.science/paper/O7EQWNH2

@misc{pith2026241218870,
  author       = {Pith},
  title        = {Pith review of: TSceneJAL: Joint Active Learning of Traffic Scenes for 3D Object Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O7EQWNH2}},
  note         = {Machine review of arXiv:2412.18870}
}
read the original abstract

Most autonomous driving (AD) datasets incur substantial costs for collection and labeling, inevitably yielding a plethora of low-quality and redundant data instances, thereby compromising performance and efficiency. Many applications in AD systems necessitate high-quality training datasets using both existing datasets and newly collected data. In this paper, we propose a traffic scene joint active learning (TSceneJAL) framework that can efficiently sample the balanced, diverse, and complex traffic scenes from both labeled and unlabeled data. The novelty of this framework is threefold: 1) a scene sampling scheme based on a category entropy, to identify scenes containing multiple object classes, thus mitigating class imbalance for the active learner; 2) a similarity sampling scheme, estimated through the directed graph representation and a marginalize kernel algorithm, to pick sparse and diverse scenes; 3) an uncertainty sampling scheme, predicted by a mixture density network, to select instances with the most unclear or complex regression outcomes for the learner. Finally, the integration of these three schemes in a joint selection strategy yields an optimal and valuable subdataset. Experiments on the KITTI, Lyft, nuScenes and SUScape datasets demonstrate that our approach outperforms existing state-of-the-art methods on 3D object detection tasks with up to 12% improvements.

Figures

Figures reproduced from arXiv: 2412.18870 by the authors.

Figure 1
Figure 1. A brief illustration of the proposed AL framework. The [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. An illustration of our TSceneJAL framework. The AL predictor [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. An illustration of graph representation of scenes and scene similarity estimation. The process comprises several steps: 1) Information [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: An illustration of the MDN and correponding uncertainy [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Experiment results of different AL methods with an increasing number of sampled scenes on the KITTI [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Number of objects in the scenes selected by different AL [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Case study of 3D detection under Random and TSceneJAL frameworks, respectively. The detector is trained on approximately 32% of the KITTI dataset. Green boxes represent true positives, red boxes represent false positives, and orange boxes indicate detections that are c…
Figure 9
Figure 9. Figure 9: An illustration of the impact of using a threshold during [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 11
Figure 11. Figure 11: Similarity distributions of the scenes selected by different [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: An illustration of the changes of perception uncertainty [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]
Figure 14
Figure 14. Figure 14: Comparison of the performances of our TSceneJAL and fully supervised learning (FSL) on the KITTI validation set. Dashed lines represent FSL results under the different difficulty levels: easy, moderate, and hard. Notably, as the dataset size reaches 80%, the detection…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 35 canonical work pages

  1. [1]

    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 Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2012, pp. 3354–3361

  2. [2]

    Lyft level 5 av dataset 2019,

    R. Kesten, M. Usman, J. Houston, T. Pandya, K. Nadhamuni, A. Ferreira, M. Yuan, B. Low, A. Jain, P. Ondruska, S. Omari, S. Shah, A. Kulkarni, A. Kazakova, C. Tao, L. Platinsky, W. Jiang, and V . Shet, “Lyft level 5 av dataset 2019,” https://level5.lyft.com/dataset/, 2019

  3. [3]

    Suscape open dataset for autonomous driving,

    SUStech ISUS Group, “Suscape open dataset for autonomous driving,” https://suscape.net/home, 2023

  4. [4]

    Sess: Self-ensembling semi- supervised 3d object detection,

    N. Zhao, T.-S. Chua, and G. H. Lee, “Sess: Self-ensembling semi- supervised 3d object detection,” in Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR) . IEEE, 2020, pp. 11 079–11 087

  5. [5]

    3dioumatch: Leveraging iou prediction for semi-supervised 3d object detection,

    H. Wang, Y . Cong, O. Litany, Y . Gao, and L. J. Guibas, “3dioumatch: Leveraging iou prediction for semi-supervised 3d object detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2021, pp. 14 615–14 624

  6. [6]

    DDS3D: Dense Pseudo-Labels with Dynamic Threshold for Semi-Supervised 3D Object Detection

    J. Li, Z. Liu, J. Hou, and D. Liang, “Dds3d: Dense pseudo-labels with dynamic threshold for semi-supervised 3d object detection,” arXiv preprint arXiv:2303.05079, 2023

  7. [7]

    Weakly supervised 3d object detection from lidar point cloud,

    Q. Meng, W. Wang, T. Zhou, J. Shen, L. Van Gool, and D. Dai, “Weakly supervised 3d object detection from lidar point cloud,” in European Conference on computer vision . Springer, 2020, pp. 515–531

  8. [8]

    Back to reality: Weakly-supervised 3d object detection with shape-guided label enhancement,

    X. Xu, Y . Wang, Y . Zheng, Y . Rao, J. Zhou, and J. Lu, “Back to reality: Weakly-supervised 3d object detection with shape-guided label enhancement,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . IEEE, 2022, pp. 8438–8447

Show all 45 references
  1. [9]

    Active learning for convolutional neural networks: A core-set approach,

    O. Sener and S. Savarese, “Active learning for convolutional neural networks: A core-set approach,” arXiv preprint arXiv:1708.00489, 2017

  2. [10]

    Deep active learning for efficient training of a lidar 3d object detector,

    D. Feng, X. Wei, L. Rosenbaum, A. Maki, and K. Dietmayer, “Deep active learning for efficient training of a lidar 3d object detector,” in2019 IEEE Intelligent Vehicles Symposium (IV) . IEEE, 2019, pp. 667–674

  3. [11]

    Deep batch active learning by diverse, uncertain gradient lower bounds,

    J. T. Ash, C. Zhang, A. Krishnamurthy, J. Langford, and A. Agar- wal, “Deep batch active learning by diverse, uncertain gradient lower bounds,” arXiv preprint arXiv:1906.03671 , 2019

  4. [12]

    Exploring active 3d object detection from a generalization perspective,

    Y . Luo, Z. Chen, Z. Wang, X. Yu, Z. Huang, and M. Baktashmotlagh, “Exploring active 3d object detection from a generalization perspective,” arXiv preprint arXiv:2301.09249 , 2023

  5. [13]

    A survey on active deep learning: from model driven to data driven,

    P. Liu, L. Wang, R. Ranjan, G. He, and L. Zhao, “A survey on active deep learning: from model driven to data driven,” ACM Computing Surveys (CSUR), vol. 54, no. 10s, pp. 1–34, 2022

  6. [14]

    Sustech points: A portable 3d point cloud interactive annotation platform system,

    E. Li, S. Wang, C. Li, D. Li, X. Wu, and Q. Hao, “Sustech points: A portable 3d point cloud interactive annotation platform system,” in 2020 IEEE Intelligent Vehicles Symposium (IV). IEEE, 2020, pp. 1108–1115

  7. [15]

    An efficient semi-automated scheme for infrastructure lidar annotation,

    A. Wu, P. He, X. Li, K. Chen, S. Ranka, and A. Rangarajan, “An efficient semi-automated scheme for infrastructure lidar annotation,” arXiv preprint arXiv:2301.10732 , 2023

  8. [16]

    Class-balanced grouping and sampling for point cloud 3d object detection,

    B. Zhu, Z. Jiang, X. Zhou, Z. Li, and G. Yu, “Class-balanced grouping and sampling for point cloud 3d object detection,” arXiv preprint arXiv:1908.09492, 2019

  9. [17]

    Revisiting domain-adaptive 3d object detection by reliable, diverse and class-balanced pseudo-labeling,

    Z. Chen, Y . Luo, Z. Wang, M. Baktashmotlagh, and Z. Huang, “Revisiting domain-adaptive 3d object detection by reliable, diverse and class-balanced pseudo-labeling,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). IEEE, 2023, pp. 3714–3726

  10. [18]

    Exploring diversity-based active learning for 3d object detection in autonomous driving,

    Z. Liang, X. Xu, S. Deng, L. Cai, T. Jiang, and K. Jia, “Exploring diversity-based active learning for 3d object detection in autonomous driving,” arXiv preprint arXiv:2205.07708 , 2022

  11. [19]

    Efficient active learning strategies for monocular 3d object detection,

    A. Hekimoglu, M. Schmidt, A. Marcos-Ramiro, and G. Rigoll, “Efficient active learning strategies for monocular 3d object detection,” in 2022 IEEE Intelligent Vehicles Symposium (IV) . IEEE, 2022, pp. 295–302

  12. [20]

    Towards complexity level classification of driving scenarios using environmental information,

    Y . Liu and J. H. Hansen, “Towards complexity level classification of driving scenarios using environmental information,” in 2019 IEEE Intelligent Transportation Systems Conference (ITSC) . IEEE, 2019, pp. 810–815

  13. [21]

    Diverse complexity measures for dataset curation in self-driving,

    A. Sadat, S. Segal, S. Casas, J. Tu, B. Yang, R. Urtasun, and E. Yumer, “Diverse complexity measures for dataset curation in self-driving,” in 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2021, pp. 8609–8616

  14. [22]

    Uncertainty-aware learning from demonstration using mixture density networks with sampling-free variance modeling,

    S. Choi, K. Lee, S. Lim, and S. Oh, “Uncertainty-aware learning from demonstration using mixture density networks with sampling-free variance modeling,” in 2018 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2018, pp. 6915–6922

  15. [23]

    Active learn- ing for deep object detection via probabilistic modeling,

    J. Choi, I. Elezi, H.-J. Lee, C. Farabet, and J. M. Alvarez, “Active learn- ing for deep object detection via probabilistic modeling,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) . IEEE, 2021, pp. 10 264–10 273

  16. [24]

    Active learning literature survey,

    A. Krishnakumar, “Active learning literature survey,” Tech. rep., Tech- nical reports, University of California, Santa Cruz. , vol. 42, 2007

  17. [25]

    Variational adversarial active learning,

    S. Sinha, S. Ebrahimi, and T. Darrell, “Variational adversarial active learning,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) . IEEE, 2019, pp. 5972–5981

  18. [26]

    Active learning with clustering,

    Z. Bod ´o, Z. Minier, and L. Csat ´o, “Active learning with clustering,” in Active Learning and Experimental Design workshop In conjunction with AISTATS 2010. JMLR Workshop and Conference Proceedings, 2011, pp. 127–139

  19. [27]

    Traffic anomaly detection model using k-means and active learning method,

    N. Liao and X. Li, “Traffic anomaly detection model using k-means and active learning method,” International Journal of Fuzzy Systems, vol. 24, no. 5, pp. 2264–2282, 2022

  20. [28]

    Dropout as a bayesian approximation: Representing model uncertainty in deep learning,

    Y . Gal and Z. Ghahramani, “Dropout as a bayesian approximation: Representing model uncertainty in deep learning,” in international conference on machine learning . PMLR, 2016, pp. 1050–1059

  21. [29]

    A new active labeling method for deep learning,

    D. Wang and Y . Shang, “A new active labeling method for deep learning,” in 2014 International Joint Conference on Neural Networks (IJCNN). IEEE, 2014, pp. 112–119

  22. [30]

    Multi-class active learning for image classification,

    A. J. Joshi, F. Porikli, and N. Papanikolopoulos, “Multi-class active learning for image classification,” in Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR) . IEEE, 2009, pp. 2372–2379

  23. [31]

    A mathematical theory of communication,

    C. E. Shannon, “A mathematical theory of communication,” ACM SIGMOBILE mobile computing and communications review , vol. 5, no. 1, pp. 3–55, 2001

  24. [32]

    Probability estimates for multi-class classification by pairwise coupling,

    T.-F. Wu, C.-J. Lin, and R. Weng, “Probability estimates for multi-class classification by pairwise coupling,” Advances in Neural Information Processing Systems, vol. 16, 2003

  25. [33]

    Reading digits in natural images with unsupervised feature learning,

    Y . Netzer, T. Wang, A. Coates, A. Bissacco, B. Wu, A. Y . Ng et al. , “Reading digits in natural images with unsupervised feature learning,” in NIPS workshop on deep learning and unsupervised feature learning . Granada, 2011, p. 4

  26. [34]

    Localization- based active learning (local) for object detection in 3d point clouds,

    A. Moses, S. Jakkampudi, C. Danner, and D. Biega, “Localization- based active learning (local) for object detection in 3d point clouds,” in Geospatial Informatics XII , vol. 12099. SPIE, 2022, pp. 44–58. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 16

  27. [35]

    Simple and scalable predictive uncertainty estimation using deep ensembles,

    B. Lakshminarayanan, A. Pritzel, and C. Blundell, “Simple and scalable predictive uncertainty estimation using deep ensembles,” Advances in Neural Information Processing Systems , vol. 30, 2017

  28. [36]

    The power of ensembles for active learning in image classification,

    W. H. Beluch, T. Genewein, A. N ¨urnberger, and J. M. K ¨ohler, “The power of ensembles for active learning in image classification,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 9368–9377

  29. [37]

    Deep similarity-based batch mode active learning with exploration- exploitation,

    C. Yin, B. Qian, S. Cao, X. Li, J. Wei, Q. Zheng, and I. Davidson, “Deep similarity-based batch mode active learning with exploration- exploitation,” in 2017 IEEE International Conference on Data Mining (ICDM). IEEE, 2017, pp. 575–584

  30. [38]

    Entropy-based active learning for object detection with progressive diversity constraint,

    J. Wu, J. Chen, and D. Huang, “Entropy-based active learning for object detection with progressive diversity constraint,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2022, pp. 9397–9406

  31. [39]

    Bayesian active learning for classification and preference learning,

    N. Houlsby, F. Husz ´ar, Z. Ghahramani, and M. Lengyel, “Bayesian active learning for classification and preference learning,” arXiv preprint arXiv:1112.5745, 2011

  32. [40]

    Marginalized kernels between labeled graphs,

    H. Kashima, K. Tsuda, and A. Inokuchi, “Marginalized kernels between labeled graphs,” in Proceedings of the 20th international conference on machine learning (ICML) , 2003, pp. 321–328

  33. [41]

    What uncertainties do we need in bayesian deep learning for computer vision?

    A. Kendall and Y . Gal, “What uncertainties do we need in bayesian deep learning for computer vision?” Advances in Neural Information Processing Systems, vol. 30, 2017

  34. [42]

    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 (CVPR). IEEE, 2019, pp. 12 697–12 705

  35. [43]

    nuscenes: A multimodal dataset for autonomous driving,

    H. Caesar, V . Bankiti, A. H. Lang, S. V ora, V . E. Liong, Q. Xu, A. Krishnan, Y . Pan, G. Baldan, and O. Beijbom, “nuscenes: A multimodal dataset for autonomous driving,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2020,...

  36. [44]

    Openpcdet: An open-source toolbox for 3d object detection from point clouds,

    O. D. Team, “Openpcdet: An open-source toolbox for 3d object detection from point clouds,” https://github.com/open-mmlab/OpenPCDet, 2020

  37. [45]

    Train in germany, test in the usa: Making 3d object detectors generalize,

    Y . Wang, X. Chen, Y . You, L. E. Li, B. Hariharan, M. Campbell, K. Q. Weinberger, and W.-L. Chao, “Train in germany, test in the usa: Making 3d object detectors generalize,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 202...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.