Pith. sign in

REVIEW 3 major objections 5 minor 53 references

StarNet: Targeted Computation for Object Detection in Point Clouds

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read StarNet shows that a LiDAR object detector can generate proposals by simply sampling point locations with zero learned parameters, featurize only local neighborhoods, and still match or beat convolutional baselines while allowing a single…

desk verdict A genuinely different point-cloud detector—sampling-based proposals, local-only featurization, adaptive compute—that works; the large Waymo margin over PointPillars is real but against a self-reimplemented baseline, so treat that number with care. read the letter →

arxiv 1908.11069 v3 pith:LKOYQ5S2 submitted 2019-08-29 cs.CV

Jiquan Ngiam , Benjamin Caine , Wei Han , Brandon Yang , Yuning Chai , Pei Sun , Yin Zhou , Xi Yi
show 5 more authors
Ouais Alsharif Patrick Nguyen Zhifeng Chen Jonathon Shlens Vijay Vasudevan
This is my paper · ORCID
classification cs.CV
keywords LiDARpointcloud3DobjectdetectiontargetedcomputationfarthestsamplinglocalfeaturizationautonomousdrivingStarNet
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 argues that the conventional design for 3D LiDAR detection—discretizing the cloud into a dense grid and running convolutions over every location—is a mismatch for sparse driving data. StarNet instead samples a modest number of point locations directly from the cloud, featurizes only the points inside a small radius around each sampled center, and predicts boxes relative to that center. The authors claim that this design, with no learned region-proposal network and no global context, is competitive with convolutional baselines on KITTI and can beat a strong convolutional baseline on Waymo pedestrian detection by more than 7 absolute mAP while costing less computation. They further claim that a single trained StarNet can be run at different compute budgets without retraining, and that seeding samples with previous-frame detections improves pedestrian mAP by roughly 40 percent relative. A sympathetic reader would care because the paper points to a different scalability path for perception: instead of making the dense grid faster, decide where the computation goes.

What carries the argument

The load-bearing component is the StarNet featurizer, built from stacked StarNet blocks. Each block aggregates the local point set with a max operation, concatenates that summary back onto every point's features, and passes the result through two fully connected layers with batch normalization and ReLU; after each block a mean readout is taken, and the readouts across blocks are concatenated into a 384-dimensional feature per proposal. Centers are chosen by sampling the point cloud's $(x,y)$ positions—random uniform, farthest point sampling, or farthest point sampling seeded with previous-frame detections—so proposals follow the data distribution. Around each center, $K$ points within radius $R$ are gathered and re-centered, and a grid of anchor offsets is placed relative to the center. This combination is what removes the need for both discretization and learned proposals, and it is what lets the number of centers and points per center be changed at inference time.

What would settle it

Train two otherwise identical StarNet models that differ only in the featurization radius $R$ (e.g., the paper's 2–3 m versus a much larger radius), and compare mAP on objects whose visible point extent is comparable to or larger than the small radius. If the larger-radius model is substantially more accurate, the sufficiency-of-local-information premise is false; if the models tie, the premise is supported.

Watch

Extended reading notes

Core claim

The paper's central claim is that cheap, parameter-free sampling of point locations is sufficient to generate detection proposals in LiDAR, and that processing each proposal from only its local point neighborhood is enough to classify and localize objects. On the Waymo Open Dataset test set, the pedestrian detector reaches 67.8 mAP with a direction-aware head against 60.0 for the authors' reimplemented PointPillars baseline, with fewer floating-point operations; vehicle results are competitive rather than dominant. The same trained pedestrian model, evaluated with different numbers of centers and points per center, traces out accuracy-versus-cost curves that sit above the baseline family across most of the operating range. Using the previous frame's high-confidence detections as 192 of 384 sample centers increases validation pedestrian mAP from 41.8 to 58.0, a relative gain of about 40 percent.

Load-bearing premise

The load-bearing premise is that the points within a small radius of a sampled center contain enough evidence to judge whether an object is there and to locate it; the paper never experiments with removing or enlarging this local neighborhood, so the claim rests on that assumption.

Editorial extensions

If this is right

  • A LiDAR detector can avoid spending computation on empty space, because proposals are sampled only where points exist; coverage results show farthest point sampling reaches high recall with a few hundred centers.
  • One trained model can be operated across a wide compute-accuracy spectrum: at similar cost to the baseline it gives about 48 percent higher pedestrian mAP, and at similar accuracy it uses about 20 percent of the baseline's FLOPs.
  • Temporal context can be added without changing the network, simply by using previous-frame detections as sample seeds; the paper reports up to about 40 percent relative mAP improvement on pedestrians.
  • The anchor grid becomes data-dependent rather than fixed, since anchor positions are defined relative to sampled centers, which lets proposals concentrate where objects actually are.

Reading between the lines

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

  • A natural extension is to replace or augment farthest point sampling with a learned importance module; the paper's own results leave open how much headroom a learned sampler would add.
  • The local-only design suggests that in dense scenes with many overlapping objects, the independence of proposals could become a limitation, so a variant that allows neighboring proposals to exchange information would directly test the cost of that independence assumption.
  • The single-model cost flexibility could be paired with a runtime policy that chooses the number of centers based on scene complexity, which the paper demonstrates is possible but does not implement.
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

3 major / 5 minor

Summary. StarNet is a LiDAR-based 3D object detector that replaces learned region proposals with parameter-free sampling of point-cloud centers (random uniform or farthest point sampling), featurizes each center's local neighborhood using a PointNet-style graph featurizer with no global context, and predicts boxes from a data-dependent anchor grid. The paper claims competitive or superior accuracy to convolutional baselines on KITTI and Waymo, demonstrates that a single trained model can trade compute for accuracy by varying the number of centers and the number of points per center, and shows that seeding proposals with previous-frame detections improves mAP without additional compute. The claims are supported by KITTI test-server results, Waymo test and validation comparisons to a reimplemented PointPillars baseline and to Multi-View Fusion, coverage analyses of the sampling strategies, and FLOPs-versus-mAP curves.

Significance. The principal idea—that cheap, zero-learned sampling on the native point cloud can serve as an effective proposal mechanism and that local-only featurization is sufficient for competitive detection—is a meaningful design contribution for autonomous-driving perception. The paper's strengths include open-sourced code, a direct coverage characterization of random and farthest-point sampling, the demonstration of inference-time accuracy/compute flexibility from a single trained model, and the temporal-context seeding mechanism. If the comparison to a stronger baseline were added, the Waymo results would be solid evidence for the 'competitive or superior' claim. The central limitations are the reliance on a self-reimplemented PointPillars baseline for the headline improvement and the absence of variance or error-bar information for any reported mAP.

major comments (3)
  1. [§4.3, Table 2, and footnote 1] The abstract's headline claim of outperforming a 'competitive baseline' on Waymo pedestrians by more than 7 absolute mAP rests solely on a comparison to the authors' own PointPillars reimplementation, which the footnote concedes is slightly weaker than the published model on KITTI (74.5 vs. 74.99 moderate). No comparison against an independently evaluated PointPillars model on Waymo is provided, and no error bars accompany the mAP differences. The reported 7.8 mAP and 12.6 mAPH advantages could therefore be partly artifacts of an under-tuned baseline. Since 'superior' is a central claim, please either add a comparison to a stronger or independently reported Waymo baseline (e.g., a third-party implementation or leaderboard result) or soften the claim to 'competitive,' while noting that Table 3's pedestrian result against Multi-View Fusion is the strongest available evidence of superiority.
  2. [§4.2 and Table 1] The KITTI test numbers are used to support the 'competitive' claim, but the text states that data augmentation changed mAP by up to +30.5 and that checkpoint selection and submission filtering (e.g., removing detections with projected height below 25 pixels) were 'extremely important.' With no error bars, multiple-seed statistics, or a described checkpoint-selection protocol, differences such as StarNet's 73.99 versus PointPillars' 74.99 on car moderate are within the range of selection noise. Please report variance or specify the selection procedure; otherwise the KITTI comparison is not quantitatively meaningful.
  3. [§4.3 and Figure 7] The computational-efficiency claim is based entirely on FLOPs, but the proposed system relies on sampling and gather operations whose FLOPs may not translate into wall-clock latency or memory savings when compared with dense convolutions. Figure 7 compares StarNet against five separately trained PointPillars models at different grid resolutions, and the claim of being 'more computationally efficient' would be substantially strengthened by reporting latency on the same hardware, or at least by specifying the exact FLOPs-counting procedure and the batch size and hardware used. Without this, the efficiency comparison is not fully established.
minor comments (5)
  1. [§4.3, footnote 1] The footnote contains a duplicated word: '74.5, 57.1, and 59.0 mAP for for cars.' Also, because this footnote is the only quantitative acknowledgement that the baseline is weaker, the information should be moved into the main text so that it is not easy to miss.
  2. [Figure 5] The coverage definition depends on 'sampled anchor boxes' with IoU > 0.5, but the anchor grid, rotations, and dimensions used for the coverage computation are not specified. Please state these parameters so the coverage numbers are reproducible.
  3. [Table 3] StarNet's vehicle mAP on the Waymo validation set is 53.7 against 62.9 for Multi-View Fusion and 57.2 for the reimplemented PointPillars. The substantial vehicle-class deficit is not discussed in the text; the paper should address this asymmetry rather than presenting it only in a table.
  4. [§4.4 and Table 4] The temporal-context result with 384 centers and 192 previous-frame detections is reported only on the validation set (41.8 to 58.0 mAP). The test-set temporal results in Table 2 are for the 1024-center configuration, where the improvement is much smaller (about 2 mAP). Reporting a test-set result for the lower-center regime would make the claimed 40% relative improvement more robust.
  5. [§3.2] The local-sufficiency premise—that a small-radius point neighborhood contains enough information for classification and localization—is never isolated experimentally. A variant with an added global context feature, or a grid-based proposal mechanism under otherwise identical settings, would make the attribution of performance to the local-only design more concrete.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: StarNet's claims are empirical evaluations against external benchmarks, with no fitted quantity renamed as a prediction.

full rationale

The paper's derivation does not define its outputs in terms of its inputs. StarNet generates proposals by zero-parameter sampling (random uniform or farthest point sampling), featurizes independent local neighborhoods with a stack of StarNet blocks, and regresses box residuals relative to data-dependent anchors; the measured mAP/mAPH numbers are then compared with PointPillars on KITTI and Waymo and with Multi-View Fusion on Waymo validation. No equation in Sections 3.1-3.4 reduces to a fitted parameter or to an assumed target value. The headline '7+ absolute mAP' advantage is an empirical test-set result against a reimplemented PointPillars baseline (acknowledged in the footnote as slightly weaker than the published model), which is a baseline-strength concern rather than a circularity. The temporal-context experiment uses the model's own prior-frame detections to seed sampling, but this is a legitimate test of the targeting mechanism, not a self-definition. The only overlapping-author citation is Multi-View Fusion [52], used purely as a comparative baseline and not as justification for any architectural choice or uniqueness claim, so self-citation is not load-bearing. The ground-truth z-percentile used for Waymo sampling is a static input-range heuristic and does not leak labels into the predicted outputs. The core claims are self-contained against external benchmarks, so no circular step is present.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claims rest on a set of standard design hyperparameters and domain assumptions about point clouds and object geometry. No new physical entities or fitted scientific constants are introduced. The most consequential assumption is that local information suffices for detection, which the paper validates empirically only indirectly.

free parameters (6)
  • number of proposals (centers) N = 1024 for main models, 64-4096 in analyses
    Controls recall and compute; chosen by validation tuning.
  • points per proposal K = 32 to 1024, typically 128-384
    Controls local context and compute; varied at inference.
  • radius R = 2-3 m
    Defines local neighborhood around each center; hand-chosen.
  • anchor grid size G = not specified exactly (G x G offsets)
    Sets the anchor density per center; standard design choice.
  • IoU thresholds for label assignment = 0.6 foreground, 0.45 background
    Determines positive/negative anchors; standard values.
  • z-range for sampling = KITTI z >= -1.35; Waymo 10th-90th percentile of object z centers
    Excludes ground points from proposal sampling; data-dependent heuristic.
assumptions (5)
  • domain assumption Every reflected point must belong to an object or surface
    Used in Section 1 to justify that sampling points as proposal centers will naturally land on objects and surfaces.
  • domain assumption 3D objects have real world scale, no perspective distortion, and rarely overlap
    Used in Section 1 to motivate local-only processing and anchor-based regression without global context.
  • domain assumption Local point clouds around sampled centers contain sufficient information for classification and regression
    Core design premise in Section 3.2; the featurizer uses only points within a small radius of each center.
  • standard math IoU overlap thresholds (0.6/0.45) produce correct foreground/background assignment
    Standard practice in anchor-based detectors; used in Section 3.3.
  • domain assumption The reimplemented PointPillars baseline is a fair proxy for convolutional baselines
    The paper compares against its own implementation, which it says is slightly weaker than the original; this assumption underlies the claimed margin of superiority.

how reviews work

0 comments
Cite this review

Pith. "Pith review of StarNet: Targeted Computation for Object Detection in Point Clouds." pith.science (2026). https://pith.science/paper/LKOYQ5S2

@misc{pith2026190811069,
  author       = {Pith},
  title        = {Pith review of: StarNet: Targeted Computation for Object Detection in Point Clouds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LKOYQ5S2}},
  note         = {Machine review of arXiv:1908.11069}
}
read the original abstract

Detecting objects from LiDAR point clouds is an important component of self-driving car technology as LiDAR provides high resolution spatial information. Previous work on point-cloud 3D object detection has re-purposed convolutional approaches from traditional camera imagery. In this work, we present an object detection system called StarNet designed specifically to take advantage of the sparse and 3D nature of point cloud data. StarNet is entirely point-based, uses no global information, has data dependent anchors, and uses sampling instead of learned region proposals. We demonstrate how this design leads to competitive or superior performance on the large Waymo Open Dataset and the KITTI detection dataset, as compared to convolutional baselines. In particular, we show how our detector can outperform a competitive baseline on Pedestrian detection on the Waymo Open Dataset by more than 7 absolute mAP while being more computationally efficient. We show how our redesign---namely using only local information and using sampling instead of learned proposals---leads to a significantly more flexible and adaptable system: we demonstrate how we can vary the computational cost of a single trained StarNet without retraining, and how we can target proposals towards areas of interest with priors and heuristics. Finally, we show how our design allows for incorporating temporal context by using detections from previous frames to target computation of the detector, which leads to further improvements in performance without additional computational cost.

Figures

Figures reproduced from arXiv: 1908.11069 by the authors.

Figure 1
Figure 1. StarNet overview. set of points are unordered and may be of arbitrary size de￾pending on the number of reflections identified by a sen￾sor on a single scan. Ideally, learned representations for point clouds aim to be permutation invariant with respect to i and agnostic to the number of points N in a given example [32, 33]. On-going efforts have attempted to design models that operate directly on point cloud data, so… view at source ↗
Figure 2
Figure 2. Example of random uniform sampling (left) and [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. StarNet point cloud featurizer. StarNet blocks are stacked, where each block’s output is read out using mean aggregation. The readouts are concatenated together to form the featurization for the point cloud. where each point has an associated feature vector. Each block first computes aggregate statistics (max) across the point cloud. Next, the global statistics are concatenated back to each point’s feature. Finally,… view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: Simple sampling procedures have good coverage over ground truth bounding boxes. The coverage of proposals for cars and vehicles is plotted against the number of samples on KITTI (left) and Waymo Open Dataset (right). Error bars (not shown) range from 0.5%-3.0%. See tex…
Figure 6
Figure 6. Figure 6: Adaptive computation with a single trained model. Waymo Open Dataset Validation set mAP on pedes￾trians of a single StarNet model trained with 1024 propos￾als, evaluated with 64 to 1024 proposals. 4. Results We present results on the KITTI object detection bench￾mark […
Figure 7
Figure 7. Figure 7: Flexible computational cost of detection for (left) pedestrians and (right) vehicles. Across 5 separately-trained PointPillars models [22], computational cost grows quadratically with increased spatial resolution for the LiDAR pseudo￾image. All curves for StarNet arise…
Figure 8
Figure 8. Figure 8: Leveraging previous proposals. Using the high￾est confidence predicted centers from the previous frames can help improve detection mAP in the next frame. based breakdowns available in Appendix A. StarNet is com￾petitive on Vehicle detection to our PointPillars [22] bas…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 41 canonical work pages

  1. [1]

    1, 4, 5, 6, 7

    Waymo open dataset: An autonomous driving dataset, 2019. 1, 4, 5, 6, 7

    work page 2019

  2. [2]

    Chauf- feurnet: Learning to drive by imitating the best and synthe- sizing the worst

    Mayank Bansal, Alex Krizhevsky, and Abhijit Ogale. Chauf- feurnet: Learning to drive by imitating the best and synthe- sizing the worst. arXiv preprint arXiv:1812.03079, 2018. 1

    arXiv 2018

  3. [3]

    Fully-convolutional siamese networks for object tracking

    Luca Bertinetto, Jack Valmadre, Joao F Henriques, Andrea Vedaldi, and Philip HS Torr. Fully-convolutional siamese networks for object tracking. In European conference on computer vision, pages 850–865. Springer, 2016. 9

    work page 2016

  4. [4]

    End to end learning for self-driving cars

    Mariusz Bojarski, Davide Del Testa, Daniel Dworakowski, Bernhard Firner, Beat Flepp, Prasoon Goyal, Lawrence D Jackel, Mathew Monfort, Urs Muller, Jiakai Zhang, et al. End to end learning for self-driving cars. arXiv preprint arXiv:1604.07316, 2016. 1, 3

    arXiv 2016

  5. [5]

    Learning to rank using gradient descent

    Christopher Burges, Tal Shaked, Erin Renshaw, Ari Lazier, Matt Deeds, Nicole Hamilton, and Gregory N Hullender. Learning to rank using gradient descent. In Proceedings of the 22nd International Conference on Machine learning (ICML-05), pages 89–96, 2005. 9

    work page 2005

  6. [6]

    Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom

    Holger Caesar, Varun Bankiti, Alex H. Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom. nuscenes: A mul- timodal dataset for autonomous driving. arXiv preprint arXiv:1903.11027, 2019. 1

    arXiv 1903

  7. [7]

    Learning to rank: from pairwise approach to listwise ap- proach

    Zhe Cao, Tao Qin, Tie-Yan Liu, Ming-Feng Tsai, and Hang Li. Learning to rank: from pairwise approach to listwise ap- proach. In Proceedings of the 24th international conference on Machine learning, pages 129–136. ACM, 2007. 9

    work page 2007

  8. [8]

    A multi-sensor fusion system for moving object detection and tracking in urban driving environments

    Hyunggi Cho, Young-Woo Seo, BVK Vijaya Kumar, and Ragunathan Raj Rajkumar. A multi-sensor fusion system for moving object detection and tracking in urban driving environments. In 2014 IEEE International Conference on Robotics and Automation (ICRA) , pages 1836–1843. IEEE,

    work page 2014

Show all 53 references
  1. [9]

    Learning to order things

    William W Cohen, Robert E Schapire, and Yoram Singer. Learning to order things. In Advances in Neural Information Processing Systems, pages 451–457, 1998. 9

    1998

  2. [10]

    Fast, accu- rate detection of 100,000 object classes on a single machine

    Thomas Dean, Mark A Ruzon, Mark Segal, Jonathon Shlens, Sudheendra Vijayanarasimhan, and Jay Yagnik. Fast, accu- rate detection of 100,000 object classes on a single machine. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1814–1821, 2013. 2

    2013

  3. [11]

    Detect to track and track to detect

    Christoph Feichtenhofer, Axel Pinz, and Andrew Zisserman. Detect to track and track to detect. In Proceedings of the IEEE International Conference on Computer Vision , pages 3038–3046, 2017. 9

    2017

  4. [12]

    Object detection with discriminatively trained part-based models

    Pedro F Felzenszwalb, Ross B Girshick, David McAllester, and Deva Ramanan. Object detection with discriminatively trained part-based models. IEEE transactions on pattern analysis and machine intelligence , 32(9):1627–1645, 2010. 2

    2010

  5. [13]

    Vision meets robotics: The kitti dataset

    Andreas Geiger, Philip Lenz, Christoph Stiller, and Raquel Urtasun. Vision meets robotics: The kitti dataset. The Inter- national Journal of Robotics Research , 32(11):1231–1237,

  6. [14]

    Fast r-cnn

    Ross Girshick. Fast r-cnn. In Proceedings of the IEEE inter- national conference on computer vision , pages 1440–1448,

  7. [15]

    Rich feature hierarchies for accurate object detection and semantic segmentation

    Ross Girshick, Jeff Donahue, Trevor Darrell, and Jitendra Malik. Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE con- ference on computer vision and pattern recognition , pages 580–587, 2014. 1, 2, 5

    2014

  8. [16]

    Re 3: Real-time recurrent regression networks for visual tracking of generic objects

    Daniel Gordon, Ali Farhadi, and Dieter Fox. Re 3: Real-time recurrent regression networks for visual tracking of generic objects. IEEE Robotics and Automation Letters , 3(2):788– 795, 2018. 9

    2018

  9. [17]

    Learn- ing to track at 100 fps with deep regression networks

    David Held, Sebastian Thrun, and Silvio Savarese. Learn- ing to track at 100 fps with deep regression networks. In European Conference on Computer Vision, pages 749–765. Springer, 2016. 9

    2016

  10. [18]

    Speed/accuracy trade-offs for modern convolutional object detectors

    Jonathan Huang, Vivek Rathod, Chen Sun, Menglong Zhu, Anoop Korattikara, Alireza Fathi, Ian Fischer, Zbigniew Wo- jna, Yang Song, Sergio Guadarrama, et al. Speed/accuracy trade-offs for modern convolutional object detectors. InPro- ceedings of the IEEE conference on computer v...

    2017

  11. [19]

    Parallel scheduling for cyber- physical systems: Analysis and case study on a self-driving car

    Junsung Kim, Hyoseung Kim, Karthik Lakshmanan, and Ragunathan Raj Rajkumar. Parallel scheduling for cyber- physical systems: Analysis and case study on a self-driving car. In Proceedings of the ACM/IEEE 4th international conference on cyber-physical systems , pages 31–40. ACM,

  12. [20]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 ,

    arXiv

  13. [21]

    Imagenet classification with deep convolutional neural net- works

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural net- works. In Advances in neural information processing sys- tems, pages 1097–1105, 2012. 2

    2012

  14. [22]

    Pointpillars: Fast en- coders for object detection from point clouds

    Alex H Lang, Sourabh V ora, Holger Caesar, Lubing Zhou, Jiong Yang, and Oscar Beijbom. Pointpillars: Fast en- coders for object detection from point clouds. arXiv preprint arXiv:1812.05784, 2018. 2, 3, 4, 5, 6, 7, 8, 12

    2018 arXiv

  15. [23]

    Cornernet: Detecting objects as paired keypoints

    Hei Law and Jia Deng. Cornernet: Detecting objects as paired keypoints. In Proceedings of the European Confer- ence on Computer Vision (ECCV), pages 734–750, 2018. 2

    2018

  16. [24]

    Deep learning

    Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. Deep learning. nature, 521(7553):436, 2015. 2

    2015

  17. [25]

    Feature pyramid networks for object detection

    Tsung-Yi Lin, Piotr Doll ´ar, Ross Girshick, Kaiming He, Bharath Hariharan, and Serge Belongie. Feature pyramid networks for object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition ,

  18. [26]

    Focal loss for dense object detection

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Doll´ar. Focal loss for dense object detection. In Pro- ceedings of the IEEE international conference on computer vision, pages 2980–2988, 2017. 1, 2, 5

    2017

  19. [27]

    Ssd: Single shot multibox detector

    Wei Liu, Dragomir Anguelov, Dumitru Erhan, Christian Szegedy, Scott Reed, Cheng-Yang Fu, and Alexander C Berg. Ssd: Single shot multibox detector. In European con- ference on computer vision, pages 21–37. Springer, 2016. 2

    2016

  20. [28]

    Fast and furious: Real time end-to-end 3d detection, tracking and motion fore- casting with a single convolutional net

    Wenjie Luo, Bin Yang, and Raquel Urtasun. Fast and furious: Real time end-to-end 3d detection, tracking and motion fore- casting with a single convolutional net. InProceedings of the IEEE Conference on Computer Vision and Pattern Recogni- tion, pages 3569–3577, 2018. 2, 3, 6

    2018

  21. [29]

    Meyer, Ankit Laddha, Eric Kee, Carlos Vallespi- Gonzalez, and Carl K

    Gregory P. Meyer, Ankit Laddha, Eric Kee, Carlos Vallespi- Gonzalez, and Carl K. Wellington. Lasernet: An effi- cient probabilistic 3d object detector for autonomous driv- ing. In The IEEE Conference on Computer Vision and Pat- tern Recognition (CVPR), June 2019. 3

    2019

  22. [30]

    Multi-task multi-sensor fusion for 3d object detec- tion

    Ming Ming Liang, Bin Yang, Yun Chen, Rui Hu, and Raquel Urtasun. Multi-task multi-sensor fusion for 3d object detec- tion. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019. 8

    2019

  23. [31]

    Frustum pointnets for 3d object detection from rgb- d data

    Charles R Qi, Wei Liu, Chenxia Wu, Hao Su, and Leonidas J Guibas. Frustum pointnets for 3d object detection from rgb- d data. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 918–927, 2018. 2, 3, 8

    2018

  24. [32]

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

    Charles R Qi, Hao Su, Kaichun Mo, and Leonidas J Guibas. Pointnet: Deep learning on point sets for 3d classification and segmentation. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 652–660,

  25. [33]

    Pointnet++: Deep hierarchical feature learning on point sets in a metric space

    Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J Guibas. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. In Advances in Neural Infor- mation Processing Systems , pages 5099–5108, 2017. 2, 3, 4

    2017

  26. [34]

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

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

    2016

  27. [35]

    Faster R-CNN: Towards real-time object detection with re- gion proposal networks

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster R-CNN: Towards real-time object detection with re- gion proposal networks. In Advances in Neural Information Processing Systems, pages 91–99, 2015. 1, 2, 9

    2015

  28. [36]

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

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. In Advances in neural information pro- cessing systems, pages 91–99, 2015. 2

    2015

  29. [37]

    Pedestrian detection with unsupervised multi-stage feature learning

    Pierre Sermanet, Koray Kavukcuoglu, Soumith Chintala, and Yann LeCun. Pedestrian detection with unsupervised multi-stage feature learning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 3626–3633, 2013. 2

    2013

  30. [38]

    PointR- CNN: 3d object proposal generation and detection from point cloud

    Shaoshuai Shi, Xiaogang Wang, and Hongsheng Li. PointR- CNN: 3d object proposal generation and detection from point cloud. In Proceedings of the IEEE Conference on Com- puter Vision and Pattern Recognition, pages 770–779, 2019. 2, 3

    2019

  31. [39]

    Stan- ley: The robot that won the darpa grand challenge

    Sebastian Thrun, Mike Montemerlo, Hendrik Dahlkamp, David Stavens, Andrei Aron, James Diebel, Philip Fong, John Gale, Morgan Halpenny, Gabriel Hoffmann, et al. Stan- ley: The robot that won the darpa grand challenge. Journal of field Robotics, 23(9):661–692, 2006. 1

    2006

  32. [40]

    Selective search for ob- ject recognition

    Jasper RR Uijlings, Koen EA Van De Sande, Theo Gev- ers, and Arnold WM Smeulders. Selective search for ob- ject recognition. International journal of computer vision , 104(2):154–171, 2013. 2

    2013

  33. [41]

    Deep parametric continu- ous convolutional neural networks

    Shenlong Wang, Simon Suo, Wei-Chiu Ma, Andrei Pokrovsky, and Raquel Urtasun. Deep parametric continu- ous convolutional neural networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recogni- tion, pages 2589–2597, 2018. 3

    2018

  34. [42]

    Pointconv: Deep convolutional networks on 3d point clouds

    Wenxuan Wu, Zhongang Qi, and Li Fuxin. Pointconv: Deep convolutional networks on 3d point clouds. arXiv preprint arXiv:1811.07246, 2018. 4

    2018 arXiv

  35. [43]

    At- tentional ShapeContextNet for point cloud recognition

    Saining Xie, Sainan Liu, Zeyu Chen, and Zhuowen Tu. At- tentional ShapeContextNet for point cloud recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 4606–4615, 2018. 8

    2018

  36. [44]

    How powerful are graph neural networks? arXiv preprint arXiv:1810.00826, 2018

    Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. How powerful are graph neural networks? arXiv preprint arXiv:1810.00826, 2018. 4

    2018 arXiv

  37. [45]

    Second: Sparsely embed- ded convolutional detection

    Yan Yan, Yuxing Mao, and Bo Li. Second: Sparsely embed- ded convolutional detection. Sensors, 18(10):3337, 2018. 2, 3, 4, 5, 6

    2018

  38. [46]

    Hdnet: Ex- ploiting HD maps for 3d object detection

    Bin Yang, Ming Liang, and Raquel Urtasun. Hdnet: Ex- ploiting HD maps for 3d object detection. In Conference on Robot Learning, pages 146–155, 2018. 2, 3, 6, 8

    2018

  39. [47]

    Pixor: Real- time 3d object detection from point clouds

    Bin Yang, Wenjie Luo, and Raquel Urtasun. Pixor: Real- time 3d object detection from point clouds. In Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 7652–7660, 2018. 2, 3, 6

    2018

  40. [48]

    Ipod: Intensive point-based object detector for point cloud

    Zetong Yang, Yanan Sun, Shu Liu, Xiaoyong Shen, and Ji- aya Jia. Ipod: Intensive point-based object detector for point cloud. arXiv preprint arXiv:1812.05276, 2018. 2, 3, 8

    2018 arXiv

  41. [49]

    Deep sets

    Manzil Zaheer, Satwik Kottur, Siamak Ravanbakhsh, Barn- abas Poczos, Ruslan R Salakhutdinov, and Alexander J Smola. Deep sets. In Advances in neural information pro- cessing systems, pages 3391–3401, 2017. 3

    2017

  42. [50]

    End-to-end inter- pretable neural motion planner

    Wenyuan Zeng, Wenjie Luo, Simon Suo, Abbas Sadat, Bin Yang, Sergio Casas, and Raquel Urtasun. End-to-end inter- pretable neural motion planner. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 8660–8669, 2019. 1

    2019

  43. [51]

    Bottom-up object detection by grouping extreme and center points

    Xingyi Zhou, Jiacheng Zhuo, and Philipp Kr ¨ahenb¨uhl. Bottom-up object detection by grouping extreme and center points. arXiv preprint arXiv:1901.08043, 2019. 2

    1901 arXiv

  44. [52]

    End-to-end multi-view fusion for 3d object detec- tion in lidar point clouds

    Yin Zhou, Pei Sun, Yu Zhang, Dragomir Anguelov, Jiyang Gao, Tom Ouyang, James Guo, Jiquan Ngiam, and Vijay Va- sudevan. End-to-end multi-view fusion for 3d object detec- tion in lidar point clouds. In Conference on Robot Learning (CoRL), 2019. 3, 5, 7, 8

    2019

  45. [53]

    V oxelnet: End-to-end learning for point cloud based 3d object detection

    Yin Zhou and Oncel Tuzel. V oxelnet: End-to-end learning for point cloud based 3d object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 4490–4499, 2018. 2, 3, 6 Supplementary Material: Targeted Computation for Object Detectio...

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.