Pith. sign in

REVIEW 3 major objections 6 minor 49 references

Crowd Counting with Deep Structured Scale Integration Network

T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read A network that refines multi-scale features through conditional random fields and trains with a dilated multi-scale structural-similarity loss achieves lower crowd-counting errors than previous methods on four benchmarks.

desk verdict Solid crowd-counting architecture with strong internal ablations, but the SOTA numbers are not independently checkable — no code, no error bars, and a small random-split benchmark makes the UCF CC 50 gain fragile. read the letter →

arxiv 1908.08692 v1 pith:PMV5CEYI submitted 2019-08-23 cs.CV

classification cs.CV
keywords crowdcountingdensitymapestimationscalevariationconditionalrandomfieldsstructuralsimilaritylossmulti-scalefeaturefusionimagepyramiddeeplearning
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 addresses crowd counting in dense scenes, where the principal obstacle is that people appear at very different sizes. It proposes a Deep Structured Scale Integration Network (DSSINet) built from three shared-parameter branches that take three scaled versions of the same image, and claims that treating the multi-scale features as variables in a conditional random field, so that each scale refines the others through message passing, makes the representation robust to scale variation. It also introduces a dilated multi-scale structural-similarity loss, which compares estimated and ground-truth density maps in local regions of several sizes, to enforce local consistency of head sizes. On four benchmarks the method reports lower mean absolute error than the previous best methods: 9.5% lower on ShanghaiTech Part A and 24.9% lower on UCF-QNRF, with fewer parameters than the main competing network. If correct, this shows that structured feature refinement and structured losses, rather than simple concatenation and pixel-wise Euclidean error, are what drive crowd-counting accuracy.

What carries the argument

The load-bearing mechanism is the Structured Feature Enhancement Module (SFEM): a continuous conditional random field in which each scale-specific feature is a random variable, the unary potential ties the refined feature to its original form, and pairwise bilinear potentials allow each scale to pass complementary information to every other scale. Mean-field inference reduces to the iterative update $h_i^t = f_i + \sum_{j \neq i} W_{ij} h_j^{t-1}$, implemented with shared $1\times1$ convolutions; the paper uses two iterations. The second mechanism is the DMS-SSIM loss, a five-layer dilated convolutional network with fixed Gaussian kernels and dilation rates 1, 2, 3, 6, 9, so that per-pixel structural similarity, a standard image-quality measure comparing luminance, contrast, and structure, is measured in regions whose receptive fields grow from $5\times5$ to $85\times85$; the loss is $1 - \prod_i \text{SSIM}(X_i,Y_i)^{\alpha_i}$.

What would settle it

Re-implement DSSINet with the published architecture and train on ShanghaiTech Part A twice, once with the DMS-SSIM loss and once with the Euclidean loss while keeping all other settings identical; if the Euclidean-trained model matches or beats the DMS-SSIM model, the claimed role of the structured loss collapses. Similarly, setting the CRF iteration count to 2 versus 3 should reproduce the reported 60.63 versus 63.80 MAE gap; failure to reproduce these two ablations would undermine the central claim.

Watch

Extended reading notes

Core claim

The paper's central claim is that a crowd-counting network can be improved on both sides of the pipeline: instead of fusing multi-scale features by concatenation or weighted averaging, the features are mutually refined by a CRF-based Structured Feature Enhancement Module, and instead of a pixel-wise Euclidean loss, training uses a Dilated Multiscale Structural Similarity (DMS-SSIM) loss. With these two components, DSSINet estimates crowd density maps whose integral gives the count, and the reported experiments on ShanghaiTech, UCF-QNRF, UCF CC 50, and WorldExpo'10 show consistent reductions in MAE and MSE over published methods. The ablation study attributes the gain specifically to the two components: removing the feature refinement raises MAE from 60.63 to 68.85 on ShanghaiTech Part A, replacing it with concatenation raises it to 67.11, and replacing the DMS-SSIM loss with a Euclidean loss raises it to 67.68.

Load-bearing premise

The comparisons assume that every method, including DSSINet, was trained and evaluated under identical preprocessing, patch sampling, and hyperparameter conditions, and with no released code this cannot be independently verified.

Editorial extensions

If this is right

  • On the four benchmarks reported, DSSINet produces lower MAE and MSE than the published methods it compares against; the headline numbers are 60.63 on ShanghaiTech Part A, 99.1 on UCF-QNRF, 216.9 on UCF CC 50, and an average of 6.67 on WorldExpo'10.
  • The CRF-based mutual refinement is doing real work: the ablations show MAE 68.85 without refinement, 67.11 with concatenation-based refinement, and 60.63 with SFEM, all on ShanghaiTech Part A.
  • The dilated multi-scale loss is doing real work: with the same network, the Euclidean loss gives 67.68, undilated MS-SSIM-5 gives 63.51, and DMS-SSIM-5 gives 60.63.
  • Two CRF iterations are optimal; a third iteration degrades MAE to 63.80, suggesting that the features become over-mixed and lose their own semantic meaning.
  • The whole network has 8.85 million parameters, roughly half of the compared CSRNet model, and the paper reports a processing time of 450 ms per 720x576 frame, which it argues is fast enough for practical surveillance use.

Reading between the lines

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

  • A direct extension is to treat the SFEM update as a linear dynamical system: because the iteration is linear, the refined features form a fixed linear combination of the original features, so one could analyze the spectral properties of the weight matrix or make the message passing input-dependent.
  • The same two-part recipe, mutual CRF refinement of multi-scale features plus a dilated multi-scale structural-similarity loss, could transfer to other dense prediction tasks with large intra-image scale variation, such as cell counting, vehicle counting, or medical lesion segmentation, where local object sizes are also correlated.
  • The ablation pattern, where the dilation schedule matters and too many scales hurt, predicts that the exact dilation schedule, not just the number of scales, is the sensitive ingredient; varying the schedule while holding the number of layers at five should produce measurable changes in MAE.
  • Because the paper reports that three CRF iterations degrade performance, a natural next step would be to learn the iteration count per image or to gate the message passing so that features that are already consistent receive less mixing.
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 / 6 minor

Summary. The paper proposes DSSINet, a deep network for crowd counting that combines three shared-parameter VGG16 subnetworks fed with a three-level image pyramid, a CRF-based Structured Feature Enhancement Module (SFEM) that mutually refines multiscale features via learned 1x1 convolutions, and a Dilated Multiscale Structural Similarity (DMS-SSIM) loss that encourages locally consistent density maps. The authors claim state-of-the-art results on four benchmarks: Shanghaitech Part A/B, UCF-QNRF, UCF CC 50, and WorldExpo'10, with relative MAE improvements of 9.5% on Shanghaitech Part A and 24.9% on UCF-QNRF over the best prior methods. Ablation studies in Section 4.4 evaluate the contribution of each component and support the design choices, and a complexity analysis in Table 9 reports parameter efficiency relative to several baselines.

Significance. If the empirical claims are substantiated, the paper makes a useful contribution to crowd counting. The SFEM is a parameter-efficient mechanism for integrating multiscale features, and the DMS-SSIM loss is an interesting adaptation of a classic image-quality metric to a high-level task. The ablations in Tables 6, 7, and 8 consistently show gains from each proposed component, and the parameter count comparison is a clear strength. However, the central claim is an empirical state-of-the-art claim that currently rests on single-run benchmark numbers without error bars, and the model-selection protocol raises concerns about the unbiasedness of the reported Shanghaitech Part A result. The paper is technically sound in its architecture and loss design, but the empirical validation needs to be made statistically and reproducibly rigorous.

major comments (3)
  1. [Section 4.4, Tables 6-8] The final configuration (CRF-2, three pyramid scales, DMS-SSIM-5) is selected by comparing test-set MAE on Shanghaitech Part A and then reported as the final Part A result in Table 2. Using the test set for model selection makes the reported 9.5% improvement over SANet optimistic and not an unbiased estimate of generalization. The authors should either use a held-out validation split for hyperparameter selection or report both validation and test results, and they should discuss the potential selection bias introduced by this protocol.
  2. [Tables 2-5] All reported numbers are single runs with no error bars or significance tests. This is especially problematic for UCF CC 50, which contains only 50 images and uses a random five-fold split; the claimed 19.1% MAE improvement over SANet could fall within split-to-split variance. I request the authors to report means and standard deviations over at least several random seeds or splits, and where feasible to include paired tests between DSSINet and the strongest baselines, to establish that the observed differences are not noise.
  3. [Section 4.1 and WorldExpo'10 experiment] The evaluation protocol is underspecified for reproduction. The paper reports patch size, optimizer, and initial learning rate, but not the number of training epochs, the learning-rate schedule, data augmentation, or the way the WorldExpo'10 Region of Interest is applied when computing MAE. Since no code or pretrained models are released, these details are necessary for an independent check of the comparison with prior baselines. Please provide a complete training and evaluation protocol, or release the code and models.
minor comments (6)
  1. [Section 3.3, Eq. (11)] The notation sigma^2_{XiYi} is used for a covariance, but the superscript 2 is misleading; use sigma_{XiYi} for the covariance term.
  2. [Section 3.2, Eq. (3)] The pairwise potential sum is written as sum_{i,j}, which implicitly includes i=j, whereas the inference update in Eq. (6) uses j does not equal i. The summation range should be made explicit and consistent.
  3. [Table 4] The caption states that the top two performances are highlighted in red and blue, but the table contains no visual highlighting. The text also says thirteen methods are compared, but the table lists nine; please align the text and table.
  4. [Section 4.4, paragraph on image-pyramid scales] The sentence 'Since the computation was too large when the scale ratio was set to 4 or larger, we did not include more' is vague; please specify the computational cost or runtime for larger scale ratios to justify the omission.
  5. [Section 3.1, Eq. (1)] The symbol w_i used for the 3x3 fusion convolutional layers collides with the pairwise weights w_i^j introduced in Section 3.2; using distinct notations would reduce confusion.
  6. [Section 3.3, Eq. (7)] The fixed-point iteration h^t_i = f_i + sum_{j does not equal i} w_i^j h^{t-1}_j is presented as mean-field inference for the CRF, but no convergence or contraction condition is given. With n=2 the update is a truncated expansion; a short derivation or a clarifying statement about this approximation would make the connection to the CRF formulation more precise.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: DSSINet's structured feature module and DMS-SSIM loss are derived from external CRF/MS-SSIM machinery, and all headline numbers are external-benchmark evaluations.

full rationale

The paper's derivation chain is self-contained against external machinery rather than reducing to its own inputs. The Structured Feature Enhancement Module is defined by a CRF energy (Eqs. 2-5) and solved by mean-field inference, yielding the message-passing update in Eqs. 6-7; the update is a learned convolutional operation whose parameters are trained, not set to reproduce the evaluation metric. The DMS-SSIM loss is a dilated reimplementation of the standard MS-SSIM loss of Wang et al. [39] with fixed Gaussian kernels and dilation rates; its weights alpha_i are taken from [39], so the loss is not fitted to the reported MAE values. Ground-truth density maps use the standard geometry-adaptive Gaussian construction of Zhang et al. [47], and the estimated count is the integral of the predicted density map, not a quantity inserted into the loss or architecture. The headline results are measured against published numbers on four external benchmarks (Tables 2-5), with no equation in the paper that takes the benchmark MAE as an input and returns it as an output. A mild evaluation caveat is that several hyperparameters (CRF iterations, pyramid scales, DMS-SSIM depth) are selected on Shanghaitech Part A in Section 4.4 and then the same Part A result is reported as headline; this is a benchmark-tuning/reproducibility concern rather than a definitional circularity, because the selected configurations are still trained and evaluated under the standard protocol and the claimed gains on UCF-QNRF, UCF CC 50, and WorldExpo'10 are not used in the selection. Absence of code and error bars affects verifiability, not circularity.

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

The central claim depends primarily on standard deep learning ingredients: pretrained VGG-16 features, Adam optimization, and benchmark-specific training targets. The fresh assumptions are the design choices for the CRF message passing and the dilated MS-SSIM loss, both justified only by the paper's own ablations.

free parameters (6)
  • Image pyramid scale ratios = 2, 1, 0.5
    Three input scales (2x, 1x, 0.5x) chosen by ablation on Shanghaitech Part A (Table 7); the '2+1+0.5' setting gives the best MAE.
  • CRF iteration count n = 2
    Selected from ablation in Table 6; n=2 yields the best MAE, and n=3 degrades performance.
  • DMS-SSIM dilation rates = [1, 2, 3, 6, 9]
    Set by hand in Section 3.3 to enlarge receptive fields; not optimized per dataset.
  • DMS-SSIM number of scales m = 5
    Chosen from Table 8; DMS-SSIM-5 gives the best result on Shanghaitech Part A.
  • DMS-SSIM Gaussian kernel sigma = 1.0
    Fixed kernel with size 5x5 and standard deviation 1.0 (Section 3.3), following standard SSIM practice.
  • Geometry-adaptive kernel spread factor = 0.3 (sigma = 0.3 * mean distance to 3 nearest neighbors)
    From prior work [47], used unchanged in Section 4.1 for all ground-truth density maps.
assumptions (4)
  • domain assumption Ground-truth density maps are generated with geometry-adaptive Gaussian kernels where sigma is 30% of the mean distance to the three nearest neighbors (Section 4.1).
    This standard protocol from [47] defines the training targets; if prior methods used different kernel settings, the reported comparisons may not be apples-to-apples.
  • domain assumption The DMS-SSIM loss, which averages luminance, contrast, and structure over multiple dilated scales, is an appropriate objective for learning accurate density maps (Section 3.3).
    The paper assumes that preserving structural similarity at multiple scales correlates with counting accuracy; this is supported only by the own ablations in Table 8, not by external analysis.
  • ad hoc to paper The fixed-point iteration in Eq. (7) with n=2 and shared 1x1 convolutions approximates the continuous CRF solution well enough to improve the features (Section 3.2).
    The convergence and optimality of this hand-designed iteration are not analyzed; the module is justified purely by the ablation results.
  • domain assumption A VGG-16 network pretrained on ImageNet provides a suitable initialization for the three shared-parameter subnetworks (Section 4.1).
    Transfer learning from ImageNet is standard, but the paper does not test alternative initializations, so the performance gain partly relies on this upstream assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Crowd Counting with Deep Structured Scale Integration Network." pith.science (2026). https://pith.science/paper/PMV5CEYI

@misc{pith2026190808692,
  author       = {Pith},
  title        = {Pith review of: Crowd Counting with Deep Structured Scale Integration Network},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PMV5CEYI}},
  note         = {Machine review of arXiv:1908.08692}
}
read the original abstract

Automatic estimation of the number of people in unconstrained crowded scenes is a challenging task and one major difficulty stems from the huge scale variation of people. In this paper, we propose a novel Deep Structured Scale Integration Network (DSSINet) for crowd counting, which addresses the scale variation of people by using structured feature representation learning and hierarchically structured loss function optimization. Unlike conventional methods which directly fuse multiple features with weighted average or concatenation, we first introduce a Structured Feature Enhancement Module based on conditional random fields (CRFs) to refine multiscale features mutually with a message passing mechanism. In this module, each scale-specific feature is considered as a continuous random variable and passes complementary information to refine the features at other scales. Second, we utilize a Dilated Multiscale Structural Similarity loss to enforce our DSSINet to learn the local correlation of people's scales within regions of various size, thus yielding high-quality density maps. Extensive experiments on four challenging benchmarks well demonstrate the effectiveness of our method. Specifically, our DSSINet achieves improvements of 9.5% error reduction on Shanghaitech dataset and 24.9% on UCF-QNRF dataset against the state-of-the-art methods.

Figures

Figures reproduced from arXiv: 1908.08692 by the authors.

Figure 1
Figure 1. Visualization of people with various scales in uncon [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the information of people’s scales on [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. The overall framework of the proposed Deep Structured Scale Integration Network (DSSINet). DSSINet consists of three [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The network of Dilated Multiscale Structural Similarity [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Visualization of the crowd density maps generated by our method on Shanghaitech Part A. The first row shows the testing images [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 41 canonical work pages

  1. [1]

    Divide and grow: Capturing huge diversity in crowd images with incrementally growing cnn

    Deepak Babu Sam, Neeraj N Sajjan, R Venkatesh Babu, and Mukundhan Srinivasan. Divide and grow: Capturing huge diversity in crowd images with incrementally growing cnn. In CVPR, pages 3618–3626, 2018

  2. [2]

    Segnet: A deep convolutional encoder-decoder architecture for image segmentation

    Vijay Badrinarayanan, Alex Kendall, and Roberto Cipolla. Segnet: A deep convolutional encoder-decoder architecture for image segmentation. arXiv preprint arXiv:1511.00561, 2015

  3. [3]

    Crowdnet: A deep convolutional network for dense crowd counting

    Lokesh Boominathan, Srinivas SS Kruthiventi, and R Venkatesh Babu. Crowdnet: A deep convolutional network for dense crowd counting. In ACM MM , pages 640–644. ACM, 2016

  4. [4]

    Scale aggregation network for accurate and efficient crowd count- ing

    Xinkun Cao, Zhipeng Wang, Yanyun Zhao, and Fei Su. Scale aggregation network for accurate and efficient crowd count- ing. In ECCV, pages 734–750, 2018

  5. [5]

    Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolu- tion, and fully connected crfs

    Liang-Chieh Chen, George Papandreou, Iasonas Kokkinos, Kevin Murphy, and Alan L Yuille. Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolu- tion, and fully connected crfs. PAMI, 40(4):834–848, 2018

  6. [6]

    Crf-cnn: Modeling structured information in human pose estimation

    Xiao Chu, Wanli Ouyang, Xiaogang Wang, et al. Crf-cnn: Modeling structured information in human pose estimation. In NIPS, pages 316–324, 2016

  7. [7]

    Pcc net: Perspective crowd counting via spatial convolutional network

    Junyu Gao, Qi Wang, and Xuelong Li. Pcc net: Perspective crowd counting via spatial convolutional network. TCSVT, 2019

  8. [8]

    Marked point processes for crowd counting

    Weina Ge and Robert T Collins. Marked point processes for crowd counting. In CVPR, pages 2913–2920. IEEE, 2009

Show all 49 references
  1. [9]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR, pages 770–778, 2016

  2. [10]

    Densely connected convolutional net- works

    Gao Huang, Zhuang Liu, Laurens Van Der Maaten, and Kil- ian Q Weinberger. Densely connected convolutional net- works. In CVPR, volume 1, page 3, 2017

  3. [11]

    Multi-source multi-scale counting in extremely dense crowd images

    Haroon Idrees, Imran Saleemi, Cody Seibert, and Mubarak Shah. Multi-source multi-scale counting in extremely dense crowd images. In CVPR, pages 2547–2554, 2013

  4. [12]

    Composition loss for counting, density map estima- tion and localization in dense crowds

    Haroon Idrees, Muhmmad Tayyab, Kishan Athrey, Dong Zhang, Somaya Al-Maadeed, Nasir Rajpoot, and Mubarak Shah. Composition loss for counting, density map estima- tion and localization in dense crowds. In ECCV, 2018

  5. [13]

    Crowd counting by adaptively fusing predictions from an image pyramid

    Di Kang and Antoni Chan. Crowd counting by adaptively fusing predictions from an image pyramid. In BMVC, 2018

  6. [14]

    Adam: A method for stochastic optimization

    Diederik Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv:1412.6980, 2014

  7. [15]

    Efficient inference in fully connected crfs with gaussian edge potentials

    Philipp Kr ¨ahenb¨uhl and Vladlen Koltun. Efficient inference in fully connected crfs with gaussian edge potentials. In Ad- vances in neural information processing systems, pages 109– 117, 2011

  8. [16]

    Conditional random fields: Probabilistic models for seg- menting and labeling sequence data

    John Lafferty, Andrew McCallum, and Fernando CN Pereira. Conditional random fields: Probabilistic models for seg- menting and labeling sequence data. 2001

  9. [17]

    Csrnet: Di- lated convolutional neural networks for understanding the highly congested scenes

    Yuhong Li, Xiaofan Zhang, and Deming Chen. Csrnet: Di- lated convolutional neural networks for understanding the highly congested scenes. In CVPR, pages 1091–1100, 2018

  10. [18]

    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 CVPR, pages 2117–2125, 2017

  11. [19]

    Decidenet: Counting varying density crowds through attention guided detection and density estimation

    Jiang Liu, Chenqiang Gao, Deyu Meng, and Alexander G Hauptmann. Decidenet: Counting varying density crowds through attention guided detection and density estimation. In CVPR, pages 5197–5206, 2018

  12. [20]

    Crowd counting using deep recurrent spatial- aware network

    Lingbo Liu, Hongjun Wang, Guanbin Li, Wanli Ouyang, and Liang Lin. Crowd counting using deep recurrent spatial- aware network. In IJCAI, 2018

  13. [21]

    Attentive crowd flow machines

    Lingbo Liu, Ruimao Zhang, Jiefeng Peng, Guanbin Li, Bowen Du, and Liang Lin. Attentive crowd flow machines. In ACM MM, pages 1553–1561. ACM, 2018

  14. [22]

    Adcrowdnet: An attention-injective deformable convolutional network for crowd understanding

    Ning Liu, Yongchao Long, Changqing Zou, Qun Niu, Li Pan, and Hefeng Wu. Adcrowdnet: An attention-injective deformable convolutional network for crowd understanding. arXiv preprint arXiv:1811.11968, 2018

  15. [23]

    Geometric and physical constraints for drone- based head plane crowd density estimation

    Weizhe Liu, Krzysztof Maciej Lis, Mathieu Salzmann, and Pascal Fua. Geometric and physical constraints for drone- based head plane crowd density estimation. In IROS, 2019

  16. [24]

    Context- aware crowd counting

    Weizhe Liu, Mathieu Salzmann, and Pascal Fua. Context- aware crowd counting. arXiv preprint arXiv:1811.10452 , 2018

  17. [25]

    Towards perspective-free object counting with deep learning

    Daniel Onoro-Rubio and Roberto J L ´opez-Sastre. Towards perspective-free object counting with deep learning. In ECCV, pages 615–629. Springer, 2016

  18. [26]

    Automatic differentiation in pytorch

    Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Al- ban Desmaison, Luca Antiga, and Adam Lerer. Automatic differentiation in pytorch. 2017

  19. [27]

    Crowd counting via multi-view scale aggre- gation networks

    Zhilin Qiu, Lingbo Liu, Guanbin Li, Qing Wang, Nong Xiao, and Liang Lin. Crowd counting via multi-view scale aggre- gation networks. In ICME, 2019

  20. [28]

    Iterative crowd counting

    Viresh Ranjan, Hieu Le, and Minh Hoai. Iterative crowd counting. In ECCV, 2018

  21. [29]

    Continuous conditional random fields for efficient regression in large fully connected graphs

    Kosta Ristovski, Vladan Radosavljevic, Slobodan Vucetic, and Zoran Obradovic. Continuous conditional random fields for efficient regression in large fully connected graphs. In AAAI, 2013

  22. [30]

    Switching convolutional neural network for crowd counting

    Deepak Babu Sam, Shiv Surya, and R Venkatesh Babu. Switching convolutional neural network for crowd counting. In CVPR, volume 1, page 6, 2017

  23. [31]

    Crowd counting via adversarial cross-scale consistency pursuit

    Zan Shen, Yi Xu, Bingbing Ni, Minsi Wang, Jianguo Hu, and Xiaokang Yang. Crowd counting via adversarial cross-scale consistency pursuit. In CVPR, pages 5245–5254, 2018

  24. [32]

    Crowd count- ing with deep negative correlation learning

    Zenglin Shi, Le Zhang, Yun Liu, Xiaofeng Cao, Yangdong Ye, Ming-Ming Cheng, and Guoyan Zheng. Crowd count- ing with deep negative correlation learning. In CVPR, pages 5382–5390, 2018

  25. [33]

    Very deep convo- lutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep convo- lutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014

  26. [34]

    Cnn-based cas- caded multi-task learning of high-level prior and density esti- mation for crowd counting

    Vishwanath A Sindagi and Vishal M Patel. Cnn-based cas- caded multi-task learning of high-level prior and density esti- mation for crowd counting. In AVSS, pages 1–6. IEEE, 2017

  27. [35]

    Generating high- quality crowd density maps using contextual pyramid cnns

    Vishwanath A Sindagi and Vishal M Patel. Generating high- quality crowd density maps using contextual pyramid cnns. In ICCV, pages 1879–1888. IEEE, 2017

  28. [36]

    Learning to count with cnn boosting

    Elad Walach and Lior Wolf. Learning to count with cnn boosting. In ECCV, pages 660–676. Springer, 2016

  29. [37]

    Di- viding and aggregating network for multi-view action recog- nition

    Dongang Wang, Wanli Ouyang, Wen Li, and Dong Xu. Di- viding and aggregating network for multi-view action recog- nition. In ECCV, pages 451–467, 2018

  30. [38]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, Eero P Simon- celli, et al. Image quality assessment: from error visibility to structural similarity. TIP, 2004

  31. [39]

    Multi- scale structural similarity for image quality assessment

    Zhou Wang, Eero P Simoncelli, and Alan C Bovik. Multi- scale structural similarity for image quality assessment. In Asilomar Conference on Signals, Systems and Computers , volume 2, pages 1398–1402. Ieee, 2003

  32. [40]

    Spatiotempo- ral modeling for crowd counting in videos

    Feng Xiong, Xingjian Shi, and Dit-Yan Yeung. Spatiotempo- ral modeling for crowd counting in videos. In ICCV. IEEE, 2017

  33. [41]

    Learning deep struc- tured multi-scale features using attention-gated crfs for con- tour prediction

    Dan Xu, Wanli Ouyang, Xavier Alameda-Pineda, Elisa Ricci, Xiaogang Wang, and Nicu Sebe. Learning deep struc- tured multi-scale features using attention-gated crfs for con- tour prediction. In NIPS, pages 3961–3970, 2017

  34. [42]

    Multi-scale convolutional neural networks for crowd counting

    Lingke Zeng, Xiangmin Xu, Bolun Cai, Suo Qiu, and Tong Zhang. Multi-scale convolutional neural networks for crowd counting. In ICIP, pages 465–469. IEEE, 2017

  35. [43]

    Cross-scene crowd counting via deep convolutional neural networks

    Cong Zhang, Hongsheng Li, Xiaogang Wang, and Xiaokang Yang. Cross-scene crowd counting via deep convolutional neural networks. In CVPR, pages 833–841, 2015

  36. [44]

    A bi-directional message passing model for salient object de- tection

    Lu Zhang, Ju Dai, Huchuan Lu, You He, and Gang Wang. A bi-directional message passing model for salient object de- tection. In CVPR, pages 1741–1750, 2018

  37. [45]

    Crowd counting via scale-adaptive convolutional neural network

    Lu Zhang, Miaojing Shi, and Qiaobo Chen. Crowd counting via scale-adaptive convolutional neural network. In WACV. IEEE, 2018

  38. [46]

    Understanding traffic density from large- scale web camera data

    Shanghang Zhang, Guanhang Wu, Joao P Costeira, and Jose MF Moura. Understanding traffic density from large- scale web camera data. In CVPR, ICME2017

  39. [47]

    Single-image crowd counting via multi-column convolutional neural network

    Yingying Zhang, Desen Zhou, Siqin Chen, Shenghua Gao, and Yi Ma. Single-image crowd counting via multi-column convolutional neural network. In CVPR, pages 589–597, 2016

  40. [48]

    Loss functions for image restoration with neural networks

    Hang Zhao, Orazio Gallo, Iuri Frosio, and Jan Kautz. Loss functions for image restoration with neural networks. TCI, 2017

  41. [49]

    Conditional random fields as re- current neural networks

    Shuai Zheng, Sadeep Jayasumana, Bernardino Romera- Paredes, Vibhav Vineet, Zhizhong Su, Dalong Du, Chang Huang, and Philip HS Torr. Conditional random fields as re- current neural networks. In ICCV, pages 1529–1537, 2015

Pith tools

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