Pith. sign in

REVIEW 2 major objections 6 minor 43 references

Occlusion-shared and Feature-separated Network for Occlusion Relationship Reasoning

T0 review · 2 major / 6 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 Sharing the occlusion cue while separating semantic feature paths improves occlusion relationship reasoning on PIOD and BSDS ownership datasets.

desk verdict The OFNet two-path architecture is a sensible, well-ablated incremental contribution, but the headline SOTA gains are less solid than the prose suggests because the key hyperparameters were selected on the test set. read the letter →

arxiv 1908.05898 v1 pith:MNYP7IC4 submitted 2019-08-16 cs.CV eess.IV

Rui Lu , Feng Xue , Menghan Zhou , Anlong Ming , Yu Zhou This is my paper · ORCID
classification cs.CVeess.IV
keywords occlusionboundarydetectionorientationestimationborderownershipbilateralfeaturemulti-ratecontextlearnerstripeconvolutionPIODdatasetBSDS
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 occlusion relationship reasoning—deciding which object is in front and where the boundary lies—needs a network that treats occlusion edge and occlusion orientation as neither fully shared nor fully separate tasks. The authors propose OFNet, which shares the occlusion cue across the two prediction paths but learns high-level semantic features in two separate streams, and adds a bilateral feature that captures context from both the foreground and background sides of a contour. This design targets two shortcomings in prior CNN-based methods: losing the shared relevance between edge and orientation by using separate networks, and losing orientation-specific context by using a single shared stream. On the PIOD and BSDS ownership datasets, OFNet outperforms the DOOBNet baseline by 4.6% and 6.1% average precision for orientation prediction. If the result holds, the practical message is that occlusion reasoning improves when the network architecture mirrors the task's shared-and-separate structure.

What carries the argument

The load-bearing machinery is the bilateral feature, defined as the contextual information on both sides of an occlusion contour that indicates which side is foreground. The Multi-rate Context Learner (MCL) extracts it by applying dilated convolutions with rates 6, 12 and 18 together with a local 1x1 convolution branch, then fuses the branches with another 1x1 convolution; this gives an edge pixel a receptive field large enough to see both occluding and occluded objects. The Bilateral Response Fusion (BRF) module concatenates this bilateral feature with the shared occlusion cue from the decoder and passes the pair through two 3x3 convolutions, producing a fused orientation map. Finally, the Occlusion Relationship Reasoning module replaces a single 3x3 kernel with orthogonal stripe convolutions (11x3 and 3x11), which correspond to decomposing a contour's tangent direction into vertical and horizontal components, so the network perceives scene context on both sides of the edge without the memory cost of a full 11x11 kernel. Together these modules implement the paper's central design rule: share the location cue, separate the semantic features.

What would settle it

Retrain OFNet and DOOBNet from the same initializations, tune the fusion ratio and stripe size on a held-out portion of the training data, then evaluate on the untouched test split; if the 64:16 ratio and 3x11 kernel no longer beat smaller variants by comparable margins, the claimed advantage is benchmark-specific.

Watch

Extended reading notes

Core claim

OFNet claims that a single-stream shared decoder is the wrong abstraction for occlusion relationship reasoning. The network keeps one occlusion cue—the decoder feature that locates the boundary—and feeds it to both the edge path and the orientation path, while the high-level features themselves are split so each path learns the semantics it needs. For orientation, OFNet introduces the bilateral feature: a Multi-rate Context Learner with parallel dilated convolutions (rates 6, 12, 18) plus a 1x1 local branch reads foreground and background regions at multiple scales, and a Bilateral Response Fusion module combines that bilateral feature with the occlusion cue at a 64:16 channel ratio. A pair of orthogonal stripe convolutions (11x3 and 3x11) then aggregates surrounding scene context along the contour's vertical and horizontal tendencies. The result is an occlusion edge map and orientation map whose joint accuracy exceeds prior systems; on PIOD the orientation AP is 0.773 versus 0.723 for DOOBNet, and on BSDS ownership it is 0.585 versus 0.501.

Load-bearing premise

The reported gains are the best of several fusion-ratio and stripe-kernel variants selected on the test set, so the load-bearing premise is that those choices (64:16 fusion ratio, 3x11 stripes) reflect a real advantage rather than overfitting the specific benchmarks.

Editorial extensions

If this is right

  • Occlusion reasoning in natural images improves when the edge and orientation paths share only the boundary-location cue rather than all high-level features.
  • Orientation prediction benefits from explicit bilateral context, which can be learned by multi-rate dilated convolution and fused with the occlusion cue.
  • Stripe convolution gives a larger effective receptive field along contour directions at modest computational cost, replacing plain 3x3 convolution in this task.
  • The architecture transfers to the small-data BSDS ownership dataset, gaining over DOOBNet, which suggests the design also helps when training data are scarce.

Reading between the lines

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

  • The paper does not test learned or adaptive fusion ratios; a gating mechanism that lets the network decide how much bilateral context to trust per pixel could be tested against the fixed 64:16 ratio.
  • The share-what-is-common, separate-what-is-specific principle could generalize to other joint prediction tasks, such as simultaneous semantic edge detection and instance segmentation, where boundary location and object identity have a similar structure.
  • The MCL's explicit local 1x1 branch alongside dilated rates implies that pure multi-scale context loses nearby contour cues; a similar local-global split could benefit other boundary-based tasks such as border ownership from stereo cues.
  • One could probe whether the 3x11 stripe advantage is tied to the image statistics of PIOD and BSDS by evaluating on rotated or scaled versions of the same images.
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

2 major / 6 minor

Summary. This paper proposes OFNet, a CNN for occlusion relationship reasoning that jointly outputs an occlusion edge map and per-pixel occlusion orientation. The architecture uses an encoder-decoder backbone with two paths: an edge path that aggregates low- and high-level cues, and an orientation path that extracts a new "bilateral feature" via a Multi-rate Context Learner (MCL), fuses it with the shared occlusion cue in a Bilateral Response Fusion (BRF) module, and reasons with orthogonal stripe convolutions. On the PIOD and BSDS ownership datasets, the authors report state-of-the-art results, exceeding the DOOBNet baseline by 4.6% and 6.1% AP respectively on the occlusion-relationship metric, and they provide ablations of the main components (Tables 3-6). The source code is released.

Significance. The paper makes a genuinely useful architectural contribution to occlusion relationship reasoning: the explicit separation of the edge and orientation paths with a shared occlusion cue addresses a design tension that single-stream methods (DOOBNet) and fully separate networks (DOC) did not resolve, and the MCL, BRF, and stripe-convolution modules are well motivated by the geometry of the task. The ablations in Tables 3-6 are systematic, and the monotonic trends (e.g., orientation AP rising from 0.703 to 0.729 as the bilateral-feature ratio increases in Table 5) give qualitative support for the design choices. The release of source code is a further strength. The central quantitative claim -- 4.6% AP gain on PIOD and 6.1% AP on BSDS ownership over DOOBNet -- is plausible but, as detailed in the major comments, its magnitude is not yet rigorously supported because of test-set hyperparameter selection, the absence of error bars, and the small BSDS test set.

major comments (2)
  1. [Sec. 4.3 (Tables 5 and 6); Sec. 4.2 (Tables 1 and 2)] The two key hyperparameters of the final model, the 64:16 fusion ratio in BRF and the 3x11 stripe kernel, are selected by comparing five variants on the PIOD test set (Tables 5 and 6), and the same test set is then used for the headline comparisons in Tables 1 and 2; no validation split, repeated-seed runs, or error bars are reported anywhere in the paper. The size of the selection effect is comparable to the headline gain: the 64:16 vs 80:16 difference is 0.007 AP (0.729 vs 0.722) and the 3x11 vs 3x9 difference is 0.017 AP (0.729 vs 0.712), against a reported 0.046 AP improvement over DOOBNet. The reported numbers are therefore the maximum of a small grid of configurations evaluated on the benchmark that was also used for model selection, so part of the claimed margin may reflect optimistic selection bias rather than the architecture itself. The authors should either choose hyperparameters on a held-out validation split and then report the corresponding test numbers, or report the mean and standard deviation over multiple random seeds for the final configuration and for DOOBNet, so that the magnitude of the improvement can be assessed.
  2. [Sec. 4.2 (Table 1, BSDS ownership)] The BSDS ownership dataset contains only 100 test images, and the claimed 6.1% AP gain over DOOBNet (0.440 to 0.501) is a point estimate from a single run. On a test set of this size, ODS/OIS/AP differences of a few points are within plausible sampling variation, and the paper gives no confidence intervals, bootstrap estimates, or per-image paired statistics. As it stands, the BSDS half of the "remarkably advances state-of-the-art" claim is not statistically assessable; the authors should add some variance estimate or a paired comparison over the 100 test images.
minor comments (6)
  1. [Sec. 3.3, Eq. (3); Sec. 4.1] Equation (3) contains a notation error: "SL(f(ai,a i))" has an undefined function f and duplicated arguments, and the value of the loss weight lambda is never given in Sec. 4.1 despite being a free parameter of the method; the authors should provide a correct expression (including, if used, the angular wrap-around in the smooth L1 term) and report lambda, along with the learning-rate schedule, batch size M, and number of iterations, which are also absent from Sec. 4.1.
  2. [Sec. 5 (Conclusion)] In the conclusion, the proposed module is referred to as "an MSL is proposed" while everywhere else it is called MCL; this typo should be corrected.
  3. [Table 3] In Table 3, the right-hand OPR columns of "Baseline(split decoder)" are numerically identical to those of "Baseline" (ODS .702, OIS .712, AP .683); if the split decoder leaves the orientation output unchanged by construction, this should be stated explicitly, otherwise the row looks like a copy-and-paste error.
  4. [Appendix A.4 (Table 7)] Table 7 shows that the 11x11 kernel slightly outperforms the selected 3x11 kernel (e.g., PIOD OPR AP 0.732 vs 0.729), and the main text should state this trade-off explicitly rather than only in the appendix, since it bears on the claim that the stripe kernel is the preferred design.
  5. [Fig. 3] Figure 3 is difficult to read: the block labels (res1 through res5, decoder output, BRF output, upsampling factors) are typeset in very small font, and the caption does not explain the meaning of different block widths and heights; a redrawn figure with a legend would substantially improve the clarity of the architecture description.
  6. [Sec. 4.2 (Tables 1 and 2)] The comparison set contains only methods published between 2015 and 2018 (SRF-OCC, DOC-HED, DOC-DMLFOV, DOOBNet); adding concurrent or more recent occlusion-reasoning methods to the comparison would make the "state-of-the-art" claim more robust.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: OFNet's SOTA claim is an externally benchmarked empirical result, and the self-citation to the authors' prior edge module is architectural reuse, not a load-bearing derivation.

full rationale

The paper's central claim is empirical: OFNet is evaluated on external benchmarks (PIOD and BSDS ownership) against published baselines, and the improvements are measured by standard ODS/OIS/AP metrics on held-out test images. There is no equation in the paper that defines a predicted quantity in terms of the same quantity it is supposed to predict; Eq. (1) and Eq. (2) define feature transformations, and Eq. (3) is a supervised loss against ground-truth edge and orientation labels. The only self-citation, [15], supplies an edge-extraction module that the paper explicitly modifies ('Different from [15], we employ an additional convolution block...'), and the main quantitative comparisons (Tables 1-2) are against DOOBNet, not against [15], so the citation is not load-bearing as a proof of the central claim. The ablation analyses in Section 4.3 select the fusion ratio (64:16) and stripe kernel (3x11) by comparing test-set performance, which raises a legitimate concern about selection bias or overfitting of hyperparameters to the benchmark; however, that is a methodological/correctness risk, not a circular derivation. No fitted parameter is renamed as a prediction, and no self-defined quantity is used to justify itself. Therefore no pattern of circularity is exhibited, and the honest finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 3 assumptions · 2 invented entities

The central claim rests on a few design hyperparameters, several chosen via test-set ablations, and on the assumption that the datasets and evaluation metrics faithfully measure occlusion reasoning. The paper introduces conceptual features (bilateral feature, occlusion cue) rather than physical entities, and these have no independent falsifiable evidence outside the paper's experiments.

free parameters (4)
  • fusion ratio of bilateral and occlusion features in BRF = 64:16
    Selected based on ablation experiments on the PIOD test set (Section 4.3, Table 5); not validated on a hold-out set.
  • stripe convolution kernel size = 3x11 and 11x3
    Selected based on ablation experiments on the PIOD test set (Section 4.3, Table 6); larger 11x11 kernel gives marginal gains with higher memory.
  • dilation rates in MCL = 6, 12, 18
    Chosen to cover multi-scale context; no ablation is provided for these values (Section 3.2.1).
  • loss weight lambda = not specified
    The loss in Eq. (3) includes a weight lambda inherited from DOOBNet [31], but its value is not stated.
assumptions (3)
  • domain assumption The ground truth annotations in PIOD and BSDS ownership accurately represent occlusion edges and orientations.
    The central evaluation depends on the correctness of these labels.
  • domain assumption The standard metrics ODS, OIS, and AP are appropriate for comparing occlusion reasoning performance.
    The paper adopts these metrics from prior work without justification; if they are flawed, the measured improvements are misleading.
  • domain assumption The left-hand rule for defining occlusion orientation is a valid representation of depth order.
    This convention is used to define ground truth orientation; it is a modeling choice, not independently verified.
invented entities (2)
  • bilateral feature
    purpose: Represent contextual information from both foreground and background sides of a contour to support orientation prediction.
    A conceptual feature introduced in this paper; its utility is demonstrated only through the reported experiments, with no external falsifiable handle.
  • occlusion cue
    purpose: Shared decoder features that indicate boundary location, used by both the edge and orientation paths.
    A term for shared features; its existence is not independently validated outside the paper's framework.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Occlusion-shared and Feature-separated Network for Occlusion Relationship Reasoning." pith.science (2026). https://pith.science/paper/MNYP7IC4

@misc{pith2026190805898,
  author       = {Pith},
  title        = {Pith review of: Occlusion-shared and Feature-separated Network for Occlusion Relationship Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MNYP7IC4}},
  note         = {Machine review of arXiv:1908.05898}
}
read the original abstract

Occlusion relationship reasoning demands closed contour to express the object, and orientation of each contour pixel to describe the order relationship between objects. Current CNN-based methods neglect two critical issues of the task: (1) simultaneous existence of the relevance and distinction for the two elements, i.e, occlusion edge and occlusion orientation; and (2) inadequate exploration to the orientation features. For the reasons above, we propose the Occlusion-shared and Feature-separated Network (OFNet). On one hand, considering the relevance between edge and orientation, two sub-networks are designed to share the occlusion cue. On the other hand, the whole network is split into two paths to learn the high-level semantic features separately. Moreover, a contextual feature for orientation prediction is extracted, which represents the bilateral cue of the foreground and background areas. The bilateral cue is then fused with the occlusion cue to precisely locate the object regions. Finally, a stripe convolution is designed to further aggregate features from surrounding scenes of the occlusion edge. The proposed OFNet remarkably advances the state-of-the-art approaches on PIOD and BSDS ownership dataset. The source code is available at https://github.com/buptlr/OFNet.

Figures

Figures reproduced from arXiv: 1908.05898 by the authors.

Figure 1
Figure 1. (a) visualization result of DOC-HED, (b) visualization [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The schematic demonstration of the high-level feature propagation process of the state-of-the-arts and ours. (a) indicates the two [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Illustration of our proposed network architecture. The [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Illustration of our proposed Multi-rate Context Learner [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Illustration of our proposed Bilateral Response Fusion [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 7
Figure 7. Figure 7: The schematic illustration of how orientation informa [PITH_FULL_IMAGE:figures/full_fig_p005_7.png]
Figure 8
Figure 8. Figure 8: OPR results on two datasets. (a) PIOD dataset (b) BSDS ownership dataset [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 9
Figure 9. Figure 9: EPR results on two datasets. ship dataset, which is difficult to train due to the small train￾ing samples, the proposed OFNet obtains the gains of 2.8% ODS, 3.7% OIS and 6.1% AP compared with the baseline DOOBNet. Specifically, our approach increases bilateral cue betw…
Figure 10
Figure 10. Figure 10: Example results on PIOD (first four rows) and BSDS ownership dataset (last four rows). [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]
Figure 11
Figure 11. Figure 11: Occlusion relationship of various approaches. The occlusion relationship (the red arrows) is represented by orientation [PITH_FULL_IMAGE:figures/full_fig_p011_11.png]
Figure 12
Figure 12. Figure 12: Edge maps of various approaches. Column 1 st: Input image. Column 2 nd − 4 th: OFNet without high-cues, OFNet without low-cues and OFNet. Column 5 th: Ground truth [PITH_FULL_IMAGE:figures/full_fig_p011_12.png]
Figure 13
Figure 13. Figure 13: Occlusion relationship of various approaches. Column [PITH_FULL_IMAGE:figures/full_fig_p012_13.png]
Figure 14
Figure 14. Figure 14: Edge maps of various approaches. Column 1 st: Input image. Column 2 nd − 7 th: conv kernel size = 3×3, 3×5, 3×7, 3×9, 3×11 and 11×11. Column 8 th: Ground truth [PITH_FULL_IMAGE:figures/full_fig_p012_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 38 canonical work pages

  1. [1]

    Detachable object detec- tion: segmentation and depth ordering from short-baseline video

    Ayvaci Alper and Soatto Stefano. Detachable object detec- tion: segmentation and depth ordering from short-baseline video. TPAMI, 34(10):1942–1951, 2012. 1

    work page 1942

  2. [2]

    Detachable object detec- tion with efficient model selection

    Alper Ayvaci and Stefano Soatto. Detachable object detec- tion with efficient model selection. In CVPR, 2011. 1

    work page 2011

  3. [3]

    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 Yuille. Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolu- tion, and fully connected crfs. TPAMI, 40(4):834–848, 2018. 2, 4

    work page 2018

  4. [4]

    Rethinking atrous convolution for semantic image segmentation

    Liang-Chieh Chen, George Papandreou, Florian Schroff, and Hartwig Adam. Rethinking atrous convolution for semantic image segmentation. CoRR, abs/1706.05587, 2017. 2

    arXiv 2017

  5. [5]

    Recovering occlusion boundaries from an image

    Hoiem Derek, Efros Alexei, and Hebert Martial. Recovering occlusion boundaries from an image. In ICCV, 2007. 1

    work page 2007

  6. [6]

    A segmentation-aware object detection model with occlusion handling

    Tianshi Gao, Benjamin Packer, and Daphne Koller. A segmentation-aware object detection model with occlusion handling. In CVPR, 2011. 1

    work page 2011

  7. [7]

    Mask r-cnn

    Kaiming He, Georgia Gkioxari, Piotr Doll ´ar, and Ross Gir- shick. Mask r-cnn. TPAMI, 2017. 1

    work page 2017

  8. [8]

    Delving deep into rectifiers: Surpassing human-level perfor- mance on imagenet classification

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Delving deep into rectifiers: Surpassing human-level perfor- mance on imagenet classification. In ICCV, 2015. 5

    2015

Show all 43 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,

  2. [10]

    Recovering occlusion boundaries from an image

    Derek Hoiem, Alexei Efros, and Martial Hebert. Recovering occlusion boundaries from an image. IJCV, 91(3):328–346,

  3. [11]

    An online learning approach to occlusion boundary detection

    Natan Jacobson, Yoav Freund, and Truong Q Nguyen. An online learning approach to occlusion boundary detection. TIP, 2012. 1

    2012

  4. [12]

    Caffe: Convolutional architecture for fast feature embedding

    Yangqing Jia, Evan Shelhamer, Jeff Donahue, Sergey Karayev, Jonathan Long, Ross Girshick, Sergio Guadarrama, and Trevor Darrell. Caffe: Convolutional architecture for fast feature embedding. volume abs/1408.5093, 2014. 5

    2014 arXiv

  5. [13]

    Wei Liu, Andrew Rabinovich, and Alexander C. Berg. Parsenet: Looking wider to see better. Computer Science ,

  6. [14]

    Fully convolutional networks for semantic segmentation

    Jonathan Long, Evan Shelhamer, and Trevor Darrell. Fully convolutional networks for semantic segmentation. TPAMI, 39(4):640–651, 2014. 2

    2014

  7. [15]

    Context-constrained accurate contour extraction for occlu- sion edge detection

    Rui Lu, Menghan Zhou, Anlong Ming, and Yu Zhou. Context-constrained accurate contour extraction for occlu- sion edge detection. In ICME, volume abs/1903.08890,

    1903 arXiv

  8. [16]

    Object-level proposals

    Jianxiang Ma, Anlong Ming, Zilong Huang, Xinggang Wang, and Yu Zhou. Object-level proposals. InICCV, 2017. 1

    2017

  9. [17]

    Occlusion edge blur: a cue to relative visual depth

    Jonathan Marshall, Christina Burbeck, Dan Ariely, Jannick Rolland, and Kevin Martin. Occlusion edge blur: a cue to relative visual depth. Journal of the Optical Society of Amer- ica A Optics Image Science & Vision , 13(4):681–8, 1996. 1

    1996

  10. [18]

    Monocular depth ordering reasoning with occlusion edge detection and couple layers inference

    Anlong Ming, Tianfu Wu, Jianxiang Ma, Fang Sun, and Yu Zhou. Monocular depth ordering reasoning with occlusion edge detection and couple layers inference. IEEE Intelligent Systems, 31(2):54–65, 2016. 1

    2016

  11. [19]

    Feedforward semantic segmentation with zoom-out features

    Mohammadreza Mostajabi, Payman Yadollahpour, and Gre- gory Shakhnarovich. Feedforward semantic segmentation with zoom-out features. In CVPR, 2015. 2

    2015

  12. [20]

    Large kernel matters – improve semantic segmen- tation by global convolutional network

    Chao Peng, Xiangyu Zhang, Gang Yu, Guiming Luo, and Jian Sun. Large kernel matters – improve semantic segmen- tation by global convolutional network. In CVPR, 2017. 2

    2017

  13. [21]

    Doc: Deep occlusion estimation from a single image

    Wang Peng and Alan Yuille. Doc: Deep occlusion estimation from a single image. In ECCV, 2016. 1, 2, 5, 6

    2016

  14. [22]

    Fig- ure/ground assignment in natural images

    Xiaofeng Ren, Charless Fowlkes, and Jitendra Malik. Fig- ure/ground assignment in natural images. In ECCV, 2006. 1, 2, 5, 6

    2006

  15. [23]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical Image Computing & Computer-assisted Intervention, 2015. 2

    2015

  16. [24]

    Probabilistic occlusion boundary detec- tion on spatio-temporal lattices

    Mehmet Emre Sargin, Luca Bertelli, Bangalore Manjunath, and Kenneth Rose. Probabilistic occlusion boundary detec- tion on spatio-temporal lattices. In ICCV, 2009. 1

    2009

  17. [25]

    Learning depth from single monocular images

    Ashutosh Saxena, Sung Chung, and Andrew Ng. Learning depth from single monocular images. In NIPS, 2005. 1

    2005

  18. [26]

    Deepcontour: A deep convolutional feature learned by positive-sharing loss for contour detection

    Wei Shen, Xinggang Wang, Yan Wang, Xiang Bai, and Zhi- jiang Zhang. Deepcontour: A deep convolutional feature learned by positive-sharing loss for contour detection. In CVPR, 2015. 1

    2015

  19. [27]

    Object skeleton extraction in natural images by fusing scale-associated deep side outputs

    Wei Shen, Kai Zhao, Yuan Jiang, Yan Wang, Zhijiang Zhang, and Xiang Bai. Object skeleton extraction in natural images by fusing scale-associated deep side outputs. InCVPR, 2016. 2

    2016

  20. [28]

    Beyond skip connections: Top-down mod- ulation for object detection

    Abhinav Shrivastava, Rahul Sukthankar, Jitendra Malik, and Abhinav Gupta. Beyond skip connections: Top-down mod- ulation for object detection. CoRR, abs/1612.06851, 2016. 2

    2016 arXiv

  21. [29]

    Occlusion boundaries from motion: Low-level detection and mid-level reasoning

    Andrew Stein and Martial Hebert. Occlusion boundaries from motion: Low-level detection and mid-level reasoning. IJCV, 82(3):325, 2009. 1

    2009

  22. [30]

    Fast 2d border ownership assignment

    Ching Teo, Cornelia Fermuller, and Yiannis Aloimonos. Fast 2d border ownership assignment. In CVPR, 2015. 2, 6

    2015

  23. [31]

    Doobnet: Deep object occlusion boundary detection from an image

    Guoxia Wang, Xiaohui Liang, and Frederick Li. Doobnet: Deep object occlusion boundary detection from an image. In ACCV, 2018. 1, 2, 4, 5, 6, 7, 8, 10

    2018

  24. [32]

    Understanding convolution for semantic segmentation

    Panqu Wang, Pengfei Chen, Ye Yuan, Ding Liu, Zehua Huang, Xiaodi Hou, and Garrison Cottrell. Understanding convolution for semantic segmentation. InIEEE Winter Con- ference on Applications of Computer Vision , 2018. 2

    2018

  25. [33]

    Deepflux for skeletons in the wild

    Yukang Wang, Yongchao Xu, Stavros Tsogkas, Xiang Bai, Sven Dickinson, and Kaleem Siddiqi. Deepflux for skeletons in the wild. In CVPR, 2019. 2

    2019

  26. [34]

    A novel multi-layer framework for tiny obstacle discovery

    Feng Xue, Anlong Ming, Menghan Zhou, and Yu Zhou. A novel multi-layer framework for tiny obstacle discovery. In ICRA, 2019. 1

    2019

  27. [35]

    Multi-scale context aggrega- tion by dilated convolutions

    Fisher Yu and Vladlen Koltun. Multi-scale context aggrega- tion by dilated convolutions. In ICLR, 2016. 3

    2016

  28. [36]

    Dilated residual networks

    Fisher Yu, Vladlen Koltun, and Thomas Funkhouser. Dilated residual networks. In CVPR, 2017. 3, 4

    2017

  29. [37]

    Monocular object instance segmentation and depth ordering with cnns

    Ziyu Zhang, Alexander Schwing, Sanja Fidler, and Raquel Urtasun. Monocular object instance segmentation and depth ordering with cnns. In ICCV, 2015. 1

    2015

  30. [38]

    Objectness-aware tracking via double-layer model

    Menghan Zhou, Jianxiang Ma, Anlong Ming, and Yu Zhou. Objectness-aware tracking via double-layer model. In ICIP,

  31. [39]

    Fusion with diffusion for robust visual tracking

    Yu Zhou, Xiang Bai, Wenyu Liu, and Longin Jan Latecki. Fusion with diffusion for robust visual tracking. In NIPS

  32. [40]

    Similarity fusion for visual tracking

    Yu Zhou, Xiang Bai, Wenyu Liu, and Longin Jan Latecki. Similarity fusion for visual tracking. IJCV, 118(3):337–363,

  33. [41]

    Learning training samples for occlusion edge detection and its application in depth ordering inference

    Yu Zhou, Jianxiang Ma, Anlong Ming, and Xiang Bai. Learning training samples for occlusion edge detection and its application in depth ordering inference. In ICPR, 2018. 1

    2018

  34. [42]

    Human action recognition with skeleton induced discriminative approximate rigid part model

    Yu Zhou and Anlong Ming. Human action recognition with skeleton induced discriminative approximate rigid part model. PRL, 2016. 1

    2016

  35. [43]

    Online multiple targets detection and tracking from mobile robot in cluttered indoor environments with depth camera

    Yu Zhou, Yinfei Yang, Yi Meng, Xiang Bai, Wenyu Liu, and Longin Jan Latecki. Online multiple targets detection and tracking from mobile robot in cluttered indoor environments with depth camera. IJPRAI, 2014. 1 A. Appendix In this appendix material, we provide full qualitative ...

    2014

Pith tools

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