Pith. sign in

REVIEW 3 major objections 3 minor 1 cited by

MVUDA: Unsupervised Domain Adaptation for Multi-view Pedestrian Detection

T0 review · 3 major / 3 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read MVUDA shows that a multi-view pedestrian detector can be adapted to a new camera rig using only unlabeled target images, eliminating the need for auxiliary labeled datasets.

desk verdict Solid UDA method with a genuinely useful pseudo-labeling trick, but the headline numbers depend on test-set threshold selection and need re-evaluation under a fixed protocol. read the letter →

arxiv 2412.04117 v1 pith:XBJ5KIEC submitted 2024-12-05 cs.CV

Erik Brorsson , Lennart Svensson , Kristofer Bengtsson , Knut {AA}kesson This is my paper · ORCID
classification cs.CV
keywords unsuperviseddomainadaptationmulti-viewpedestriandetectionmeanteacherself-trainingpseudo-labelingbird's-eyeviewoccupancymapcross-camerarigWildtrackMultiviewX
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

MVUDA tries to establish that a multi-view pedestrian detector trained on labeled images from one camera rig can be adapted to a different rig using only unlabeled images from the target rig, with no auxiliary labeled monocular data. The paper claims this is the first strict unsupervised domain adaptation (UDA) method for multi-view pedestrian detection that eliminates the need for external labeled datasets. The method combines mean teacher self-training with a pseudo-labeling step that only accepts detections at local maxima of the predicted bird's-eye-view occupancy map. On the Wildtrack→MultiviewX benchmark, MVUDA reports raising MODA from 35.9 to 82.4, outperforming the prior UDA method [28] even though that method relied on a monocular detector trained on large labeled datasets. If the claims hold, multi-view pedestrian detection becomes practical to deploy on new camera setups without collecting target-domain annotations.

What carries the argument

The method builds on the multi-view detector of [43], a variant of [20]: a ResNet-18 (a standard convolutional image-feature extractor) produces per-view features, a perspective transformation projects these into a shared bird's-eye-view (BEV) grid, average pooling fuses the views, and dilated convolutions regress a probabilistic occupancy map (a per-cell score of pedestrian presence on the ground plane). The mean teacher [40] is an exponential moving average of the student weights; it generates pseudo-labels on unaugmented target images while the student trains on strongly augmented images, using a Gaussian-softened MSE loss on both source labels and target pseudo-labels. The novel component is local-max pseudo-labeling: a candidate detection is accepted only if its score is at least as large as every score in a surrounding square neighborhood defined by $k_d$ (a 70 cm by 70 cm square when $k_d=3$) and also exceeds a threshold $\tau$, so non-maximum points are never turned into pseudo-labels. Dropview [43] and 3D random occlusion [32] augmentations are used in both the baseline and self-training.

What would settle it

Run the two cross-domain benchmarks with $\tau$ fixed to 0.4 for both pseudo-label generation and final evaluation, without any test-set sweep, and check whether MVUDA still reaches the reported 82.4 MODA on Wildtrack→MultiviewX and 85.4 on MultiviewX→Wildtrack; the paper's own ablation shows local-max self-training ranging from 73.2 at $\tau=0.2$ to 43.5 at $\tau=0.4$ on the former, so a fixed threshold would materially change the headline result.

Watch

Extended reading notes

Core claim

The central claim is that strict unsupervised domain adaptation is achievable for multi-view pedestrian detection without any external labeled datasets or pretrained monocular detectors. The paper argues that mean teacher self-training works well when pseudo-labels are restricted to local maxima of the teacher's predicted occupancy map: vanilla threshold-and-NMS post-processing can select detections that are not local maxima, and these are unreliable exactly when the confidence threshold is ambiguous in a new domain. The proposed local-max pseudo-labeling yields reliable pseudo-labels at lower confidence levels, which matters most when the source-only baseline transfers poorly, as in Wildtrack→MultiviewX where the baseline scores 35.9 MODA and MVUDA reports 82.4. The paper also introduces two cross-scene camera-rig benchmarks (GMVD1→MultiviewX and GMVD2→MultiviewX) and reports the highest MODA among methods that do not use auxiliary labeled data on all studied benchmarks.

Load-bearing premise

The headline numbers depend on selecting the detection and pseudo-label thresholds from the target-domain test labels (the paper sweeps $\tau$ and keeps the value with the highest MODA detection-accuracy score, then sets per-benchmark $\tau$ from the final results table), so the reported gains assume the practitioner can peek at the test set before fixing thresholds.

Editorial extensions

If this is right

  • A multi-view pedestrian detector can be adapted to a new camera rig with only unlabeled target video, removing the need for expensive multi-view annotation or for an external monocular detector.
  • The largest gains occur when the source-only model is weak: on Wildtrack→MultiviewX, MVUDA raises MODA from 35.9 to 82.4, so self-training can largely rescue a detector that transfers poorly across rigs.
  • Local-max pseudo-labeling makes self-training more robust to the choice of confidence threshold, outperforming vanilla pseudo-labeling on most benchmarks and improving over baseline over a wider range of $\tau$.
  • On the new cross-scene benchmarks GMVD1→MultiviewX (89.0 MODA) and GMVD2→MultiviewX (88.8), MVUDA approaches the Oracle trained with target labels (91.2), indicating that most of the benefit of target labels is recoverable without them.
  • The framework establishes a UDA baseline that does not depend on any auxiliary labeled dataset, which the paper argues is important for practical applications and for extending to new object types.

Reading between the lines

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

  • Editorial inference: The reported numbers should be read as upper bounds because the paper selects the detection threshold by sweeping $\tau$ over {0.05, ..., 0.95} on the test set and keeping the best MODA, and it sets the self-training pseudo-label threshold per benchmark from the final results table.
  • Editorial inference: A conservative rerun with $\tau$ fixed a priori (or chosen on a held-out validation split) would likely shrink the headline gains on some benchmarks, since the paper's no-augmentation ablation shows Wildtrack→MultiviewX local-max self-training varying from 73.2 MODA at $\tau=0.2$ to 43.5 at $\tau=0.4$.
  • Editorial inference: Local-max pseudo-labeling is a generic idea that could transfer to other dense-prediction UDA problems where the model outputs a confidence map expected to peak at object centers, such as keypoint detection or semantic segmentation.
  • Editorial inference: Because the method removes the dependence on external labeled datasets, a natural testable extension is applying the same pipeline to other object classes (e.g., vehicles) using a source rig and a different target rig, where no monocular pretrained detector exists.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The paper proposes MVUDA, an unsupervised domain adaptation method for multi-view pedestrian detection. It uses a mean-teacher self-training framework in which a student detector is trained on labeled source data and pseudo-labeled target data, with pseudo-labels generated by a teacher and post-processed by a newly proposed 'local-max' procedure that retains only local maxima of the predicted occupancy map above a confidence threshold. Experiments on Wildtrack, MultiviewX, and GMVD report large gains over the source-only baseline, close-to-oracle performance on several benchmarks, and state-of-the-art results among methods that do not use auxiliary labeled data. The paper also introduces two new cross-scene benchmark splits, GMVD1→MultiviewX and GMVD2→MultiviewX.

Significance. If the reported numbers hold under a valid evaluation protocol, this is a useful contribution: it removes the dependence on external monocular detectors that characterizes prior UDA work by Lima et al., and it proposes a simple, well-motivated pseudo-label post-processing idea that appears to improve self-training. The paper is unusually thorough on ablations: Tables 3, 4, 6-8 and supplementary Tables 9-13 cover the main components (mean teacher, augmentation, pseudo-labeling variants, α, λ, kd, training length). The qualitative claim that self-training plus local-max pseudo-labeling helps is supported by these ablations. However, the quantitative headline claims are currently entangled with test-set threshold selection: both the evaluation threshold and the pseudo-label threshold are chosen using target-domain ground truth. Therefore the magnitude of the claimed improvement over prior methods is not yet established.

major comments (3)
  1. [Sec. 4.2] The evaluation protocol selects the detection threshold on the target test set: 'we evaluate the model on the range τ ∈ {0.05, 0.10, ..., 0.95} and select the result with highest MODA.' This is an oracle selection on test labels. Because prior methods are evaluated at fixed τ=0.4 (as the paper itself notes), the comparisons in Tables 1 and 2 are not on equal terms. The Wildtrack→MultiviewX headline jump from 35.9 to 82.4 is reported at τ=0.2; Table 4 shows that without augmentation the same method at τ=0.4 gives 43.5 and at τ=0.3 gives 68.7. Please re-evaluate all methods under a fixed, pre-specified threshold (at least τ=0.4) and, additionally or alternatively, choose τ without using target labels (e.g., on a labeled source validation set).
  2. [Sec. 4.5 and Table 4] The pseudo-label threshold is also chosen per benchmark from the final MODA table: the text states the threshold is set 'following the experiments presented in Tab. 4,' and Table 4 is computed from target test predictions. Since the pseudo-label threshold directly controls the training targets for self-training, the entire pipeline is tuned to the target test ground truth. This is a second use of test labels for model selection. Please either fix the pseudo-label thresholds a priori (e.g., using source-domain validation statistics) or report a sensitivity analysis showing that the conclusions are stable across a range of reasonable thresholds.
  3. [Sec. 4.5, Table 4] The claim that local-max pseudo-labeling is 'robust' to the threshold τ is not fully supported by the data. On Wildtrack→MultiviewX, local-max pseudo-labeling yields 73.2 MODA at τ=0.2, 68.7 at τ=0.3, and 43.5 at τ=0.4. A drop of ~30 MODA over 0.2 units of threshold is a strong sensitivity, not robustness. The text also says the method 'yields improvements over the baseline for a wider range of τ,' but on this benchmark the range of τ for which the improvement is large is narrow. Please qualify this claim with the actual range and report results at fixed thresholds.
minor comments (3)
  1. [Table 4 vs Table 2] The benchmark name is inconsistent: Table 2 lists 'MultiviewX 1,2,6→3,4,5' while Table 4 lists 'MultiviewX 1,2,3→4,5,6'. Please align the camera indices.
  2. [Supplementary, Sec. 5.3] The text says λ=1.0 yields 'the best overall performance (tied with λ=2.0),' but in Table 10 λ=1.0 gives 79.7 vs λ=2.0 giving 79.1 on MultiviewX→Wildtrack, and 87.8 vs 88.4 on GMVD1→MultiviewX. Please clarify which metric supports the 'best overall' statement.
  3. [Sec. 4.2] The paper states that the purpose of selecting the best MODA over τ is to 'ensure that the experimental results are not affected by the specific choice of τ.' In fact, this procedure makes the results depend on the test set through the chosen threshold; a sentence acknowledging this distinction would be more accurate.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: MVUDA is an empirical self-training method evaluated on external benchmarks; the oracle-threshold evaluation protocol is a test-set tuning caveat, not a circular step.

full rationale

The paper is an empirical self-training study rather than a formal derivation or first-principles prediction. The student model is trained on labeled source data and pseudo-labels from a mean teacher, and every reported MODA/MODP/precision/recall number is measured against held-out ground truth on external benchmarks (Wildtrack, MultiviewX, GMVD subsets). No load-bearing self-citation chain appears: the only same-author reference is [44], cited for a background remark on robot control, and no uniqueness theorem or ansatz is imported from prior work by the authors to force the model choice. The proposed local-max pseudo-labeling is a post-processing rule tested against target labels; it is not defined in terms of the final metric. The only circularity-adjacent behavior is the evaluation protocol: Sec. 4.2 sweeps the detection threshold tau over {0.05,...,0.95} and selects the highest MODA, and Sec. 4.5 sets per-benchmark pseudo-label thresholds using Table 4, which is generated from target-domain test predictions. This is test-set threshold tuning that inflates the headline comparison (e.g., Wildtrack→MultiviewX reported as 82.4 at tau=0.2, while Table 4 shows 68.7 at tau=0.3 and 43.5 at tau=0.4 for local-max pseudo-labeling without augmentation), and it means the SOTA margin over Lima et al., who use fixed tau=0.4, is not on equal terms. However, this is an evaluation-protocol weakness rather than a circular derivation: the model predictions are still measured against external ground truth, and the method improves over the 35.9 baseline even at fixed thresholds (e.g., 68.7 at tau=0.3). Therefore, no prediction reduces by construction to a fitted parameter or to a self-citation chain.

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

The central claims rest on standard supervised self-training assumptions plus a handful of hand-set hyperparameters. The method introduces no new physical entities; the only invented objects are two dataset-split benchmarks (GMVD1 to MultiviewX and GMVD2 to MultiviewX), which are evaluation protocols rather than entities. The main hidden costs are ImageNet pretraining and test-set-informed threshold selection.

free parameters (5)
  • Pseudo-label threshold tau per benchmark = 0.4 for MultiviewX to Wildtrack, 0.2 for Wildtrack to MultiviewX, 0.3 for all others
    Set to maximize final MODA in Tab. 4, not fixed a priori; directly controls which pseudo-labels enter training.
  • Evaluation threshold tau_eval = Selected from sweep 0.05 to 0.95 as argmax MODA
    Sec. 4.2: 'select the result with highest MODA'; this uses test-set labels to choose the reported operating point.
  • Local-max neighborhood size kd = 3 (70x70 cm)
    Sec. 3.3 and Tab. 6; kd=3 is chosen as the largest value fitting within the 0.5 m NMS radius, and performance varies with kd (e.g., 63.4 to 81.2 on MultiviewX to Wildtrack).
  • EMA decay alpha = 0.99 (or 0.999 for extended training)
    Sec. 4.5 Tab. 7; performance varies with alpha and training duration, with alpha=0.999 benefiting from longer training.
  • Target loss weight lambda = 1.0
    Supplementary Tab. 10; chosen as convenient, with performance tied by lambda=2.0 on one benchmark.
assumptions (3)
  • domain assumption A well-trained multi-view detector produces one local maximum in the BEV occupancy map per pedestrian after MSE training on Gaussian targets.
    Used in Sec. 3.3 to justify local-max pseudo-labeling; if false, the filter would discard true detections and reduce recall.
  • domain assumption Known camera calibration and flat ground plane allow perspective transformation to BEV; pedestrian head position uses a fixed height.
    Sec. 3.1 and Supplementary Sec. 5.4, inherited from MVDet; calibration errors or non-flat ground would corrupt BEV features and pseudo-labels.
  • domain assumption Self-training with pseudo-labels from a mean teacher converges to the target distribution without external regularization.
    Core of the method; the paper gives empirical evidence (Tabs. 1-3) but no theoretical guarantee, and low alpha causes stability issues (Sec. 4.5).

how reviews work

0 comments
Cite this review

Pith. "Pith review of MVUDA: Unsupervised Domain Adaptation for Multi-view Pedestrian Detection." pith.science (2026). https://pith.science/paper/XBJ5KIEC

@misc{pith2026241204117,
  author       = {Pith},
  title        = {Pith review of: MVUDA: Unsupervised Domain Adaptation for Multi-view Pedestrian Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XBJ5KIEC}},
  note         = {Machine review of arXiv:2412.04117}
}
abstract

We address multi-view pedestrian detection in a setting where labeled data is collected using a multi-camera setup different from the one used for testing. While recent multi-view pedestrian detectors perform well on the camera rig used for training, their performance declines when applied to a different setup. To facilitate seamless deployment across varied camera rigs, we propose an unsupervised domain adaptation (UDA) method that adapts the model to new rigs without requiring additional labeled data. Specifically, we leverage the mean teacher self-training framework with a novel pseudo-labeling technique tailored to multi-view pedestrian detection. This method achieves state-of-the-art performance on multiple benchmarks, including MultiviewX$\rightarrow$Wildtrack. Unlike previous methods, our approach eliminates the need for external labeled monocular datasets, thereby reducing reliance on labeled data. Extensive evaluations demonstrate the effectiveness of our method and validate key design choices. By enabling robust adaptation across camera setups, our work enhances the practicality of multi-view pedestrian detectors and establishes a strong UDA baseline for future research.

Figures

Figures reproduced from arXiv: 2412.04117 by the authors.

Figure 1
Figure 1. Since labeled multi-view datasets are scarce, current [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. An overview of our proposed self-training method for UDA multi-view pedestrian detection. A student is trained with labels on [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Illustrative example of predicted occupancy scores in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Example of the regressed foot heat map yˆ n f for the first camera (n = 1) in the MultiviewX dataset [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: Example of projected pseudo-label y n f for the first cam￾era (n = 1) in the MultiviewX dataset. Here, LMSE denotes the MSE-loss with a Gaussian kernel G as in the main paper, according to LMSE(y, yˆ) = XH i=1 XW j=1 (G(yij ) − yˆij ) 2 . (7) Following [20], the total …
Figure 6
Figure 6. Figure 6: A test sample from Wildtrack (a), as well as the associated label and predictions of the baseline and MVUDA (b). The predictions [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: A test sample from MultiViewX (a), as well as the associated label and predictions of the baseline and MVUDA (b). The [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. DCHM: Depth-Consistent Human Modeling for Multiview Detection

    cs.CV 2025-07 conditional novelty 6.0 of 10

    DCHM uses superpixel-based Gaussian Splatting to make monocular depth estimates multiview-consistent, producing point clouds that yield state-of-the-art label-free pedestrian detection on Wildtrack, Terrace, and MultiviewX.

Reference graph

Works this paper leans on

45 extracted references · 38 canonical work pages · cited by 1 Pith paper

  1. [1]

    Sparsity driven people localization with a heterogeneous network of cameras

    Alexandre Alahi, Laurent Jacques, Yannick Boursier, and Pierre Vandergheynst. Sparsity driven people localization with a heterogeneous network of cameras. Journal of Math- ematical Imaging and Vision, 41:39–58, 2011. 2

    work page 2011

  2. [2]

    Enhancing multi-view pedestrian detection through generalized 3D feature pulling

    Sithu Aung, Haesol Park, Hyungjoo Jung, and Junghyun Cho. Enhancing multi-view pedestrian detection through generalized 3D feature pulling. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 1196–1205, 2024. 1, 2, 6

    work page 2024

  3. [3]

    Deep oc- clusion reasoning for multi-camera multi-target detection

    Pierre Baqu ´e, Franc ¸ois Fleuret, and Pascal Fua. Deep oc- clusion reasoning for multi-camera multi-target detection. In Proceedings of the IEEE International Conference on Com- puter Vision, pages 271–279, 2017. 2

    work page 2017

  4. [4]

    Exploring object relation in mean teacher for cross-domain detection

    Qi Cai, Yingwei Pan, Chong-Wah Ngo, Xinmei Tian, Lingyu Duan, and Ting Yao. Exploring object relation in mean teacher for cross-domain detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 11457–11466, 2019. 2

    work page 2019

  5. [5]

    Contrastive mean teacher for domain adaptive ob- ject detectors

    Shengcao Cao, Dhiraj Joshi, Liang-Yan Gui, and Yu-Xiong Wang. Contrastive mean teacher for domain adaptive ob- ject detectors. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 23839– 23848, 2023. 2, 3

    work page 2023

  6. [6]

    Deep multi- camera people detection

    Tatjana Chavdarova and Franc ¸ois Fleuret. Deep multi- camera people detection. In 2017 16th IEEE international conference on machine learning and applications (ICMLA), pages 848–853. IEEE, 2017. 2

    work page 2017

  7. [7]

    Wildtrack: A multi-camera HD dataset for dense unscripted pedes- trian detection

    Tatjana Chavdarova, Pierre Baqu ´e, St ´ephane Bouquet, An- drii Maksai, Cijo Jose, Timur Bagautdinov, Louis Lettry, Pascal Fua, Luc Van Gool, and Franc ¸ois Fleuret. Wildtrack: A multi-camera HD dataset for dense unscripted pedes- trian detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 5030–5039,

    work page

  8. [8]

    Multi-camera object de- tection for robotics

    Adam Coates and Andrew Y Ng. Multi-camera object de- tection for robotics. In 2010 IEEE International conference on robotics and automation, pages 412–419. IEEE, 2010. 1

    work page 2010

Show all 45 references
  1. [9]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. IEEE, 2009. 5

    2009

  2. [10]

    Un- biased mean teacher for cross-domain object detection

    Jinhong Deng, Wen Li, Yuhua Chen, and Lixin Duan. Un- biased mean teacher for cross-domain object detection. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 4091–4101, 2021. 2

    2021

  3. [11]

    Two-level data augmentation for calibrated multi-view de- tection

    Martin Engilberge, Haixin Shi, Zhiye Wang, and Pascal Fua. Two-level data augmentation for calibrated multi-view de- tection. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages 128–136, 2023. 1, 2, 8

    2023

  4. [12]

    Pets2009: Dataset and challenge

    James Ferryman and Ali Shahrokni. Pets2009: Dataset and challenge. In 2009 Twelfth IEEE international workshop on performance evaluation of tracking and surveillance , pages 1–6. IEEE, 2009. 1

    2009

  5. [13]

    Multicamera people tracking with a probabilistic occupancy map

    Francois Fleuret, Jerome Berclaz, Richard Lengagne, and Pascal Fua. Multicamera people tracking with a probabilistic occupancy map. IEEE transactions on pattern analysis and machine intelligence, 30(2):267–282, 2007. 2

    2007

  6. [14]

    Domain-adversarial training of neural networks

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pas- cal Germain, Hugo Larochelle, Franc ¸ois Laviolette, Mario March, and Victor Lempitsky. Domain-adversarial training of neural networks. Journal of machine learning research , 17(59):1–35, 2016. 2

    2016

  7. [15]

    Dlow: Domain flow for adaptation and generalization

    Rui Gong, Wen Li, Yuhua Chen, and Luc Van Gool. Dlow: Domain flow for adaptation and generalization. In Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2477–2486, 2019. 2

    2019

  8. [16]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 3, 5

    2016

  9. [17]

    Fcns in the wild: Pixel-level adversarial and constraint-based adaptation

    Judy Hoffman, Dequan Wang, Fisher Yu, and Trevor Darrell. Fcns in the wild: Pixel-level adversarial and constraint-based adaptation. arXiv preprint arXiv:1612.02649, 2016. 2

    2016 arXiv

  10. [18]

    Cycada: Cycle-consistent adversarial domain adaptation

    Judy Hoffman, Eric Tzeng, Taesung Park, Jun-Yan Zhu, Phillip Isola, Kate Saenko, Alexei Efros, and Trevor Darrell. Cycada: Cycle-consistent adversarial domain adaptation. In International conference on machine learning, pages 1989–

    1989

  11. [19]

    Multiview detection with shadow transformer (and view-coherent data augmentation)

    Yunzhong Hou and Liang Zheng. Multiview detection with shadow transformer (and view-coherent data augmentation). In Proceedings of the 29th ACM International Conference on Multimedia, pages 1673–1682, 2021. 1, 2, 4

    2021

  12. [20]

    Multi- view detection with feature perspective transformation

    Yunzhong Hou, Liang Zheng, and Stephen Gould. Multi- view detection with feature perspective transformation. In Computer Vision – ECCV 2020 , pages 1–18, Cham, 2020. Springer International Publishing. 1, 2, 3, 4, 5

    2020

  13. [21]

    Daformer: Improving network architectures and training strategies for domain-adaptive semantic segmentation

    Lukas Hoyer, Dengxin Dai, and Luc Van Gool. Daformer: Improving network architectures and training strategies for domain-adaptive semantic segmentation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9924–9935, 2022. 2

    2022

  14. [22]

    Framework for performance evaluation of face, text, and vehicle detection and tracking in video: Data, metrics, and protocol

    Rangachar Kasturi, Dmitry Goldgof, Padmanabhan Soundararajan, Vasant Manohar, John Garofolo, Rachel Bowers, Matthew Boonstra, Valentina Korzhova, and Jing Zhang. Framework for performance evaluation of face, text, and vehicle detection and tracking in video: Data, metrics, and...

    2008

  15. [23]

    Pseudo-Label: The simple and effi- cient semi-supervised learning method for deep neural net- works

    Dong-Hyun Lee et al. Pseudo-Label: The simple and effi- cient semi-supervised learning method for deep neural net- works. In Workshop on challenges in representation learn- ing, ICML, page 896. Atlanta, 2013. 2

    2013

  16. [24]

    Multi- view target transformation for pedestrian detection

    Wei-Yu Lee, Ljubomir Jovanov, and Wilfried Philips. Multi- view target transformation for pedestrian detection. In Pro- ceedings of the IEEE/CVF Winter Conference on Applica- tions of Computer Vision, pages 90–99, 2023. 2

    2023

  17. [25]

    Cross-domain adaptive teacher for object detection

    Yu-Jhe Li, Xiaoliang Dai, Chih-Yao Ma, Yen-Cheng Liu, Kan Chen, Bichen Wu, Zijian He, Kris Kitani, and Peter Va- jda. Cross-domain adaptive teacher for object detection. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition , pages 7581–7590, 2...

    2022

  18. [26]

    3D pedestrian localization using mul- tiple cameras: A generalizable approach

    Jo ˜ao Paulo Lima, Rafael Roberto, Lucas Figueiredo, Fran- cisco Sim ˜oes, Diego Thomas, Hideaki Uchiyama, and Veronica Teichrieb. 3D pedestrian localization using mul- tiple cameras: A generalizable approach. Machine Vision and Applications, 33(4):61, 2022. 2, 3

    2022

  19. [27]

    Toward unlabeled multi-view 3D pedes- trian detection by generalizable AI: techniques and perfor- mance analysis

    Jo ˜ao Paulo Lima, Diego Thomas, Hideaki Uchiyama, and Veronica Teichrieb. Toward unlabeled multi-view 3D pedes- trian detection by generalizable AI: techniques and perfor- mance analysis. In 2023 36th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI) , pages 1–6...

    2023

  20. [28]

    Mean teacher for unsupervised domain adaptation in multi-view 3D pedestrian detection

    Jo ˜ao Paulo Lima, Diego Thomas, Hideaki Uchiyama, and Veronica Teichrieb. Mean teacher for unsupervised domain adaptation in multi-view 3D pedestrian detection. In 2024 37th SIBGRAPI Conference on Graphics, Patterns and Im- ages (SIBGRAPI), pages 1–6. IEEE, 2024. 2, 3, 6

    2024

  21. [29]

    Learning transferable features with deep adaptation net- works

    Mingsheng Long, Yue Cao, Jianmin Wang, and Michael Jor- dan. Learning transferable features with deep adaptation net- works. In International conference on machine learning , pages 97–105. PMLR, 2015. 2

    2015

  22. [30]

    Semantic-driven multi- camera pedestrian detection

    Alejandro L ´opez-Cifuentes, Marcos Escudero-Vi˜nolo, Jes´us Besc´os, and Pablo Carballeira. Semantic-driven multi- camera pedestrian detection. Knowledge and Information Systems, 64(5):1211–1237, 2022. 2

    2022

  23. [31]

    Robust multiple cameras pedestrian detection with multi-view bayesian network

    Peixi Peng, Yonghong Tian, Yaowei Wang, Jia Li, and Tiejun Huang. Robust multiple cameras pedestrian detection with multi-view bayesian network. Pattern Recognition, 48(5): 1760–1772, 2015. 2

    2015

  24. [32]

    3D random occlusion and multi-layer projection for deep multi-camera pedestrian localization

    Rui Qiu, Ming Xu, Yuyao Yan, Jeremy S Smith, and Xi Yang. 3D random occlusion and multi-layer projection for deep multi-camera pedestrian localization. In European Conference on Computer Vision , pages 695–710. Springer,

  25. [33]

    A deep top-down framework to- wards generalisable multi-view pedestrian detection

    Rui Qiu, Ming Xu, Yuchen Ling, Jeremy S Smith, Yuyao Yan, and Xinheng Wang. A deep top-down framework to- wards generalisable multi-view pedestrian detection. Neuro- computing, 607:128458, 2024. 6

    2024

  26. [34]

    PPM: A boolean optimizer for data association in multi-view pedestrian detection

    Rui Qiu, Ming Xu, Yuyao Yan, Jeremy S Smith, and Yuchen Ling. PPM: A boolean optimizer for data association in multi-view pedestrian detection. Pattern Recognition, 156: 110807, 2024. 2, 6

    2024

  27. [35]

    Multi-camera video surveillance for real-time analysis and reconstruction of soccer games

    Jinchang Ren, Ming Xu, James Orwell, and Graeme A Jones. Multi-camera video surveillance for real-time analysis and reconstruction of soccer games. Machine Vision and Appli- cations, 21:855–863, 2010. 1

    2010

  28. [36]

    Conditional random fields for multi-camera object de- tection

    Gemma Roig, Xavier Boix, Horesh Ben Shitrit, and Pascal Fua. Conditional random fields for multi-camera object de- tection. In 2011 International Conference on Computer Vi- sion, pages 563–570. IEEE, 2011. 2

    2011

  29. [37]

    Maximum classifier discrepancy for unsuper- vised domain adaptation

    Kuniaki Saito, Kohei Watanabe, Yoshitaka Ushiku, and Tat- suya Harada. Maximum classifier discrepancy for unsuper- vised domain adaptation. In Proceedings of the IEEE con- ference on computer vision and pattern recognition , pages 3723–3732, 2018. 2

    2018

  30. [38]

    Super-convergence: Very fast training of neural networks using large learn- ing rates

    Leslie N Smith and Nicholay Topin. Super-convergence: Very fast training of neural networks using large learn- ing rates. In Artificial intelligence and machine learning for multi-domain operations applications , pages 369–386. SPIE, 2019. 5

    2019

  31. [39]

    Stacked homography transformations for multi-view pedestrian detection

    Liangchen Song, Jialian Wu, Ming Yang, Qian Zhang, Yuan Li, and Junsong Yuan. Stacked homography transformations for multi-view pedestrian detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 6049–6057, 2021. 1, 2

    2021

  32. [40]

    Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results

    Antti Tarvainen and Harri Valpola. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. Advances in neural information processing systems, 30, 2017. 2

    2017

  33. [41]

    EarlyBird: Early-fusion for multi- view tracking in the bird’s eye view

    Torben Teepe, Philipp Wolters, Johannes Gilg, Fabian Her- zog, and Gerhard Rigoll. EarlyBird: Early-fusion for multi- view tracking in the bird’s eye view. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 102–111, 2024. 2

    2024

  34. [42]

    Learning to adapt structured output space for semantic seg- mentation

    Yi-Hsuan Tsai, Wei-Chih Hung, Samuel Schulter, Ki- hyuk Sohn, Ming-Hsuan Yang, and Manmohan Chandraker. Learning to adapt structured output space for semantic seg- mentation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 7472–7481,

  35. [43]

    Bringing generalization to deep multi-view pedestrian detection

    Jeet V ora, Swetanjal Dutta, Kanishk Jain, Shyamgopal Karthik, and Vineet Gandhi. Bringing generalization to deep multi-view pedestrian detection. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 110–119, 2023. 1, 2, 3, 4, 5, 6, 8

    2023

  36. [44]

    Prescient collision-free navigation of mobile robots with iterative multimodal motion prediction of dy- namic obstacles

    Ze Zhang, Hadi Hajieghrary, Emmanuel Dean, and Knut ˚Akesson. Prescient collision-free navigation of mobile robots with iterative multimodal motion prediction of dy- namic obstacles. IEEE Robotics and Automation Letters , 8 (9):5488–5495, 2023. 1

    2023

  37. [45]

    Rectifying pseudo label learn- ing via uncertainty estimation for domain adaptive seman- tic segmentation

    Zhedong Zheng and Yi Yang. Rectifying pseudo label learn- ing via uncertainty estimation for domain adaptive seman- tic segmentation. International Journal of Computer Vision, 129(4):1106–1120, 2021. 3 10 MVUDA: Unsupervised Domain Adaptation for Multi-view Pedestrian Detectio...

    2021

Pith tools

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