Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

Open-World Amodal Appearance Completion

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Text plus image reconstructs hidden objects across any category

desk verdict A genuinely new integration for text-query amodal completion with a reasonable human study, but the occlusion-ordering step is the least-evidenced link in the chain. read the letter →

arxiv 2411.13019 v1 pith:MFCPPBYD submitted 2024-11-20 cs.CV

Jiayang Ao , Yanbei Jiang , Qiuhong Ke , Krista A. Ehinger This is my paper · ORCID
classification cs.CV
keywords amodalcompletionopen-worldreasoningtext-guidedsegmentationocclusionorderinginpaintingRGBAoutputhumanpreferencestudy
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 tries to show that a user can take an ordinary photo, type a phrase naming any object—specific like 'polar bear' or abstract like 'what is the mammal in this image'—and receive a complete, transparent RGBA version of that object with its occluded parts filled in, with no additional training. The central claim is that segmentation, occlusion ordering, prompt selection, and iterative inpainting can be chained from off-the-shelf components to achieve this in the open world, beyond any fixed category set. If the claim holds, amodal completion stops being a closed-set recognition problem and becomes a language-driven image reconstruction service usable for editing, 3D reconstruction, and augmented reality. The paper supports the claim with a 2,379-image, 553-class evaluation set and a human preference study in which its outputs are chosen 41.86% of the time overall against 27.95% for the strongest baseline.

What carries the argument

The load-bearing mechanism is the occluder mask $M_{\text{occ}}$, built by InstaOrderNet's pairwise occlusion decisions over every object mask and morphological background segment (Eq. 2), with boundary-aware dilation when the target touches the image edge (Eq. 3). This mask tells the inpainting model which pixels are hidden by other things and therefore must be reconstructed. The other carrier is the prompt-selection step (Eq. 4), which uses CLIP to choose the descriptor from the image tags or the user query that best matches the visible target, so abstract queries still yield a concrete inpainting condition. The pipeline then iterates masked inpainting with adaptive termination.

What would settle it

Take the pipeline's 2,379 evaluation images, run InstaOrderNet alone over the same segment inventory the pipeline uses, and compare its pairwise occlusion labels with human judgments; if its accuracy on novel classes or background segments is near chance, then the occluder mask is initialized from wrong pixels and the claimed completions cannot be attributed to correct occlusion reasoning.

Watch

Extended reading notes

Core claim

The paper's core discovery is that reasoning-based amodal completion can be decomposed into four frozen components that need no training: a vision-language model turns the text query into a visible mask for the target object; an open-set tagging and detection chain inventories every nameable object while morphological operations partition ambiguous background into segments; an occlusion-ordering network assigns each segment a binary 'occludes the target' label to form the occluder mask; and an inpainting diffusion model, guided by a CLIP-selected prompt, iteratively fills the masked region and outputs an RGBA layer. The authors argue that because the query, rather than a predefined class list, selects the target, the framework generalizes across 553 object classes and produces completions that human raters prefer to existing closed-set and diffusion-based baselines. The central output is not just an inpainted image but a layerable transparent element ready for downstream composition.

Load-bearing premise

The whole result rests on the occlusion-ordering network's binary pairwise decisions being trustworthy for every segment the pipeline feeds it, including masks from open-vocabulary detectors and morphological background blobs, even though the paper never reports how often those decisions are correct on its own segment inventory.

Editorial extensions

If this is right

  • A text query plus an image becomes enough to recover a full amodal appearance for objects from categories never seen in training.
  • Abstract or context-dependent queries, such as 'the mammal in this image', produce completions without requiring the user to know the class name.
  • RGBA outputs let the completed object be composited directly into image editing, novel-view synthesis, and 3D reconstruction workflows.
  • Because each module is frozen and replaceable, the whole framework improves automatically when better segmentation, occlusion-ordering, or inpainting models appear.
  • The new dataset and preference protocol provide a common benchmark for open-world amodal appearance completion.

Reading between the lines

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

  • Editorial inference: the paper never measures InstaOrderNet's own accuracy on its segment inventory, so a natural extension is to ablate the pipeline by replacing the occlusion-order network with human-annotated occlusion labels; the gap would reveal how much of the final quality rests on that one component.
  • Editorial inference: the same chain could be applied to video frames, but temporal consistency of completed appearances would need an explicit constraint that the current single-image pipeline does not provide.
  • Editorial inference: completions could serve as pseudo-ground-truth for distilling a smaller, faster end-to-end model, since the current engine runs several large pre-trained models per object.
  • Editorial inference: the morphological handling of unlabeled background regions is what lets the method treat foliage and blur as occluders, a setting where closed-set baselines fail; a targeted stress test on heavy-background occlusion would make this contribution stand out.
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 / 5 minor

Summary. The paper proposes a training-free framework for open-world amodal appearance completion. Given an image and a natural-language query, the pipeline uses LISA to obtain a visible mask, Grounded-SAM plus morphological background partitioning to enumerate candidate segments, InstaOrderNet to determine which segments occlude the target (Eq. 2), CLIP to select an inpainting prompt (Eq. 4), and Stable Diffusion inpainting with iterative refinement to synthesize the occluded appearance, finally outputting an RGBA layer. The authors introduce a new evaluation dataset of 2379 images spanning 553 classes and report a human preference study (Table 2), quantitative metrics (Table 3), and ablations (Table 4).

Significance. If the results hold, this is a useful step toward open-world amodal completion: it moves beyond closed category sets, integrates off-the-shelf models without fine-tuning, and produces compositable RGBA outputs. The new benchmark with natural occlusions and the human-preference evaluation are valuable assets. However, the central empirical claim rests on a preference study without statistical inference, and the occlusion-analysis module is not validated on the pipeline's own segment inventory; these issues need to be addressed before the claims are fully supported.

major comments (3)
  1. [§4.3, Table 2] The headline result that 'Ours' is preferred (41.86% overall vs. 27.95% for Pix2gestalt) is reported without confidence intervals or significance tests. Given that each image appears to be rated by only three participants (per Fig. 5), the 13.9-point gap could be within noise for a subset of images; the paper should provide per-image bootstrap or subject-level confidence intervals and a paired test (e.g., Wilcoxon across images).
  2. [§3.2, Eq. (2)] The occluder mask is built from binary decisions of InstaOrderNet applied to every segment in S∪B, including masks produced by Grounded-SAM and morphological background blobs. No accuracy or agreement measure is reported for InstaOrderNet on this segment inventory, and there is no oracle ablation replacing it with ground-truth occlusion order. Because an incorrect binary decision either leaves an occluder unpainted or masks visible target pixels that then get hallucinated, the open-world claim is only as strong as this unverified transfer. The supplementary failure analysis (Sec. 7) attributes 4.1% complete failures to undetectable objects, not to occlusion-order errors, so this failure mode is not disentangled.
  3. [§3.3 vs. §4.2] The CLIP score metric in Table 3 is partly circular. The evaluation uses the ground-truth class label as the query Q, and Eq. (4) selects the inpainting prompt from T∪{Q} with CLIP; the same label is then matched against the output with CLIP. This inflates the reported CLIP scores, particularly for 'Ours', and makes the cross-method comparison on this metric difficult to interpret. A fairer protocol would blind prompt selection to the evaluation label or use a distinct text encoder for evaluation.
minor comments (5)
  1. [§3.1, Eq. (1)] The morphological operations (erosion and dilation structuring element sizes, number of passes) are not specified; please report them in the supplementary configuration details for reproducibility.
  2. [§4.3] The human study should state how many participants rated each image, whether each image was rated by the same set of three participants, and whether the text query was shown to participants; Fig. 5 implies three raters per image, but this is not stated in the main text.
  3. [Table 4] The symbols '!' and '#' are not defined in the table caption; the row 'T ∪ Q #' is ambiguous about whether background segments are included or excluded.
  4. [§3.1 and §4.4] There are several typos: 'an segmentation process' should be 'a segmentation process', 'our method of consider' should be 'our method to consider', and 'evalution' should be 'evaluation' in §4.4.
  5. [§4.4] The sentence 'In our evalution dataset, the Q prompt achieves the highest CLIP score' should explicitly point to Table 4 and clarify that the T∪Q row with background segments is the chosen configuration despite having a slightly lower CLIP score.

Circularity Check

1 steps flagged · score 4.0 of 10

CLIP 'class relevance' is partly circular: the ground-truth label is fed in as the query (Eq. 4) and inpainting prompt (Eq. 6), then the same label is the CLIP evaluation target; the paper concedes this in Sec. 4.4. Human preference and pixel metrics remain independent.

  1. fitted input called prediction [Sec. 4.2 'Evaluation metrics' with Sec. 3.3 Eq. (4) and Sec. 3.4 Eq. (6); conceded in Sec. 4.4]
    "We use CLIP score [26] to measure how well the amodal completions align with the class labels. Specifically, we compare the amodal completion of each object with the ground-truth class label associated with the object. ... In our evalution dataset, the Q prompt achieves the highest CLIP score, because it directly aligns with ground-truth target object class labels."

    The evaluation dataset supplies a ground-truth class label for each occluded object, and that label is used as the language query Q. Eq. (4) selects the inpainting prompt P from T ∪ {Q}, so the ground-truth label is an allowed prompt; Eq. (6) then conditions the diffusion inpainting on P. The CLIP metric in Sec. 4.2 scores the resulting completion against the same ground-truth label. Thus the 'class relevance' numbers in Table 3 partly measure whether the generator followed a prompt that already contained the answer, not independent open-world generalization. The paper concedes this in Sec. 4.4, saying the Q-only prompt wins the CLIP score because it directly aligns with the ground-truth labels.

full rationale

The framework itself is not circular: it takes an image and a text query and produces an RGBA completion through frozen components (LISA mask, Grounded-SAM with SAM segments, morphological background partition, InstaOrderNet pairwise occlusion, CLIP prompt selection, and Stable Diffusion inpainting). None of the pipeline equations define the output in terms of the CLIP evaluation target; Eq. (2) is a union of InstaOrderNet binary occlusion decisions and Eq. (6) is conditional inpainting. The unverified transfer of InstaOrderNet to open-vocabulary and background segments is a generalization risk, not a circularity, since no oracle ablation or per-module accuracy is reported; that concern belongs under correctness risk. The only concrete circular step is evaluative: the dataset's ground-truth class label is fed into the pipeline as Q via Eq. (4), the selected P conditions the inpainting in Eq. (6), and the CLIP metric in Sec. 4.2 scores the completion against that same label. The paper's own Sec. 4.4 remark that the Q prompt achieves the highest CLIP score 'because it directly aligns with ground-truth target object class labels' acknowledges this bias. Human preference in Table 2 and LPIPS/SSIM in Table 3 provide independent evidence not affected by this step, and the only self-citation ([1]) is a background survey, not load-bearing. Therefore the circularity is partial and confined to one supporting metric, yielding a score of 4.

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

Everything load-bearing is inherited from frozen pretrained models or from hand-set evaluation choices. There are no fitted coefficients in the pipeline, but the method's success depends on several hand-set parameters whose values are mostly unreported. The comparison depends on the assumptions that human preference and visible-region similarity metrics measure completion quality. No new physical or theoretical entities are introduced; 'reasoning amodal completion' names a task rather than a mechanism.

free parameters (5)
  • epsilon (occluder mask stability threshold) = not reported
    Eq. 7 stops iteration when the L1 change in Mocc falls below epsilon; the value is chosen by hand and determines how many inpainting passes run.
  • maximum inpainting iterations T = 3
    Section 4.2 limits inpainting to at most three iterations per object; this hand-set cap affects quality and computational cost.
  • dilation structuring element for boundary-aware occlusion = not reported
    Eq. 3 uses dilation d(Mvisible) along contacted edges; the kernel size is not specified.
  • morphological kernel for background segmentation = not reported
    Eq. 1 partitions unsegmented regions with erosion and dilation; the structuring element is unspecified.
  • alpha blend transition width = not reported
    Eq. 8 blends with a transition region of predefined width; the width is not given.
assumptions (6)
  • domain assumption Pretrained models (LISA, GroundingDINO, SAM, RAM++, InstaOrderNet, Stable Diffusion) provide sufficient world knowledge to segment, reason, and inpaint arbitrary open-world objects without task-specific training.
    Sections 3.1 to 3.4 and 4.2 rely on these frozen models; the framework contributes no learned weights.
  • domain assumption LISA's text-grounded mask correctly isolates the visible region of the queried object.
    Mvisible is used as ground truth for all subsequent occlusion and inpainting steps in Section 3.1.
  • domain assumption InstaOrderNet's pairwise occlusion decisions transfer to masks produced by open-vocabulary detectors and to morphological background segments.
    Eq. 2 converts each occ=1 decision into the occluder mask; the paper gives no evidence that background blobs are within InstaOrderNet's input distribution.
  • domain assumption CLIP similarity between the masked visible object and candidate tags selects the correct inpainting prompt.
    Eq. 4 chooses P as argmax CLIP(Itarget, ti); a wrong choice conditions Stable Diffusion on an incorrect appearance description.
  • domain assumption Human preference on Prolific with gold-standard checks is a reliable proxy for amodal completion quality.
    The headline comparison in Table 2 is subjective because ground-truth occluded appearance does not exist; Fleiss' kappa of 0.319 is only fair agreement.
  • domain assumption Visible-region reference metrics (LPIPS, SSIM, VGG feature similarity) are meaningful quality proxies for amodal completion.
    Section 4.2 says these metrics are provided for reference only; they compare only visible parts and can favor outputs that barely change visible pixels.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Open-World Amodal Appearance Completion." pith.science (2026). https://pith.science/paper/MFCPPBYD

@misc{pith2026241113019,
  author       = {Pith},
  title        = {Pith review of: Open-World Amodal Appearance Completion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MFCPPBYD}},
  note         = {Machine review of arXiv:2411.13019}
}
read the original abstract

Understanding and reconstructing occluded objects is a challenging problem, especially in open-world scenarios where categories and contexts are diverse and unpredictable. Traditional methods, however, are typically restricted to closed sets of object categories, limiting their use in complex, open-world scenes. We introduce Open-World Amodal Appearance Completion, a training-free framework that expands amodal completion capabilities by accepting flexible text queries as input. Our approach generalizes to arbitrary objects specified by both direct terms and abstract queries. We term this capability reasoning amodal completion, where the system reconstructs the full appearance of the queried object based on the provided image and language query. Our framework unifies segmentation, occlusion analysis, and inpainting to handle complex occlusions and generates completed objects as RGBA elements, enabling seamless integration into applications such as 3D reconstruction and image editing. Extensive evaluations demonstrate the effectiveness of our approach in generalizing to novel objects and occlusions, establishing a new benchmark for amodal completion in open-world settings. The code and datasets will be released after paper acceptance.

Figures

Figures reproduced from arXiv: 2411.13019 by the authors.

Figure 1
Figure 1. Examples of our open-world amodal completion using both specific (e.g., “polar bear”) and abstract (e.g., “What is the mammal in [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of our framework. Starting with a text query, a VLM generates a visible mask to locate the target object in the input [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Distribution of the top 50 most frequent categories in the our evaluation dataset. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Visual comparisons of amodal completions across different methods: Ours consistently outperforms others in terms of realism, [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Model preference of human evaluators by agreement [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Amodal completion results from our method, with and [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Comparison of our amodal completion results with [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Visual comparisons of amodal completions across different methods. Our method consistently outperforms others in reconstruc [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Examples of our common partial failures, such as unnatural poses and hand reconstruction. “Fails” refers in cases where no [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: A detailed guide provided to participants at the begin [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 13
Figure 13. Figure 13: Image distribution across different datasets. [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]
Figure 12
Figure 12. Figure 12: Examples of filtered images from COCO-A [ [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. SynergyAmodal: Deocclude Anything with Text Control

    cs.CV 2025-04 conditional novelty 6.0 of 10

    SynergyAmodal co-synthesizes a 16K amodal dataset from EntitySeg images with human and model guidance, and trains a diffusion model that completes occluded objects with optional text control.

  2. CAPTURe: Evaluating Spatial Reasoning in Vision Language Models via Occluded Object Counting

    cs.CV 2025-04 conditional novelty 6.0 of 10

    CAPTURe, a new benchmark for occluded pattern counting, shows that six vision-language models count far worse when objects are hidden, while humans make almost no errors.

Reference graph

Works this paper leans on

44 extracted references · 38 canonical work pages · cited by 2 Pith papers

  1. [1]

    Image amodal completion: A survey

    Jiayang Ao, Qiuhong Ke, and Krista A Ehinger. Image amodal completion: A survey. Computer Vision and Image Understanding, 229:103661, 2023. 2

    work page 2023

  2. [2]

    Text2live: Text-driven layered image and video editing

    Omer Bar-Tal, Dolev Ofri-Amar, Rafail Fridman, Yoni Kas- ten, and Tali Dekel. Text2live: Text-driven layered image and video editing. In European Conference on Computer Vision, pages 707–723. Springer, 2022. 1

    work page 2022

  3. [3]

    In- structpix2pix: Learning to follow image editing instructions

    Tim Brooks, Aleksander Holynski, and Alexei A Efros. In- structpix2pix: Learning to follow image editing instructions. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18392–18402, 2023. 1

    2023

  4. [4]

    Segan: Segmenting and generating the invisible

    Kiana Ehsani, Roozbeh Mottaghi, and Ali Farhadi. Segan: Segmenting and generating the invisible. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 6144–6153, 2018. 5

    work page 2018

  5. [5]

    Learning to see the invisible: End-to-end trainable amodal instance segmen- tation

    Patrick Follmann, Rebecca K ¨onig, Philipp H ¨artinger, Michael Klostermann, and Tobias B ¨ottger. Learning to see the invisible: End-to-end trainable amodal instance segmen- tation. In 2019 IEEE Winter Conference on Applications of Computer Vision, pages 1328–1336. IEEE, 2019. 5

    work page 2019

  6. [6]

    Im- age style transfer using convolutional neural networks

    Leon A Gatys, Alexander S Ecker, and Matthias Bethge. Im- age style transfer using convolutional neural networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 2414–2423, 2016. 6

    work page 2016

  7. [7]

    Panodr: Spherical panorama diminished reality for indoor scenes

    Vasileios Gkitsas, Vladimiros Sterzentsenko, Nikolaos Zioulis, Georgios Albanis, and Dimitrios Zarpalas. Panodr: Spherical panorama diminished reality for indoor scenes. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 3716–3726, 2021. 1

    work page 2021

  8. [8]

    Language-grounded dy- namic scene graphs for interactive object search with mobile manipulation

    Daniel Honerkamp, Martin B ¨uchner, Fabien Despinoy, Tim Welschehold, and Abhinav Valada. Language-grounded dy- namic scene graphs for interactive object search with mobile manipulation. IEEE Robotics and Automation Letters, 2024. 1

    work page 2024

Show all 44 references
  1. [9]

    Sail-vos: Semantic amodal in- stance level video object segmentation-a synthetic dataset and baselines

    Yuan-Ting Hu, Hong-Shuo Chen, Kexin Hui, Jia-Bin Huang, and Alexander G Schwing. Sail-vos: Semantic amodal in- stance level video object segmentation-a synthetic dataset and baselines. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages...

    2019

  2. [10]

    Open-set image tagging with multi-grained text su- pervision

    Xinyu Huang, Yi-Jie Huang, Youcai Zhang, Weiwei Tian, Rui Feng, Yuejie Zhang, Yanchun Xie, Yaqian Li, and Lei Zhang. Open-set image tagging with multi-grained text su- pervision. arXiv e-prints, pages arXiv–2310, 2023. 4, 6

    2023

  3. [11]

    Organi- zation in vision: Essays on gestalt perception

    Gaetano Kanizsa, Paolo Legrenzi, and Paolo Bozzi. Organi- zation in vision: Essays on gestalt perception. 1979. 1

    1979

  4. [12]

    Segment any- thing

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C Berg, Wan-Yen Lo, et al. Segment any- thing. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 4015–4026, 2023. 4, 6

    2023

  5. [13]

    Visual genome: Connecting language and vision using crowdsourced dense image annotations

    Ranjay Krishna, Yuke Zhu, Oliver Groth, Justin Johnson, Kenji Hata, Joshua Kravitz, Stephanie Chen, Yannis Kalan- tidis, Li-Jia Li, David A Shamma, et al. Visual genome: Connecting language and vision using crowdsourced dense image annotations. International Journal of Compute...

    2017

  6. [14]

    Lisa: Reasoning segmentation 6 via large language model

    Xin Lai, Zhuotao Tian, Yukang Chen, Yanwei Li, Yuhui Yuan, Shu Liu, and Jiaya Jia. Lisa: Reasoning segmentation 6 via large language model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 9579–9589, 2024. 2, 3, 6

    2024

  7. [15]

    The measurement of observer agreement for categorical data

    J Richard Landis and Gary G Koch. The measurement of observer agreement for categorical data. biometrics, pages 159–174, 1977. 4

    1977

  8. [16]

    Instance-wise occlusion and depth orders in natural scenes

    Hyunmin Lee and Jaesik Park. Instance-wise occlusion and depth orders in natural scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 21210–21221, 2022. 4, 6

    2022

  9. [17]

    2d amodal instance segmentation guided by 3d shape prior

    Zhixuan Li, Weining Ye, Tingting Jiang, and Tiejun Huang. 2d amodal instance segmentation guided by 3d shape prior. In European Conference on Computer Vision , pages 165–

  10. [18]

    Muva: A new large-scale benchmark for multi-view amodal instance segmentation in the shopping scenario

    Zhixuan Li, Weining Ye, Juan Terven, Zachary Bennett, Ying Zheng, Tingting Jiang, and Tiejun Huang. Muva: A new large-scale benchmark for multi-view amodal instance segmentation in the shopping scenario. In Proceedings of the IEEE/CVF International Conference on Computer Vi- s...

    2023

  11. [19]

    Variational amodal object completion

    Huan Ling, David Acuna, Karsten Kreis, Seung Wook Kim, and Sanja Fidler. Variational amodal object completion. Advances in Neural Information Processing Systems , 33: 16246–16257, 2020. 1

    2020

  12. [20]

    Grounding dino: Marrying dino with grounded pre-training for open-set object detection

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Qing Jiang, Chunyuan Li, Jianwei Yang, Hang Su, et al. Grounding dino: Marrying dino with grounded pre-training for open-set object detection. In Eu- ropean Conference on Computer Vision. Springer, 2024. 4, 6

    2024

  13. [21]

    Object-level scene deocclusion

    Zhengzhe Liu, Qing Liu, Chirui Chang, Jianming Zhang, Daniil Pakhomov, Haitian Zheng, Zhe Lin, Daniel Cohen-Or, and Chi-Wing Fu. Object-level scene deocclusion. In ACM SIGGRAPH 2024 Conference Papers, pages 1–11, 2024. 2

    2024

  14. [22]

    Towards open- world interactive disambiguation for robotic grasping

    Yuchen Mo, Hanbo Zhang, and Tao Kong. Towards open- world interactive disambiguation for robotic grasping. In 2023 IEEE International Conference on Robotics and Au- tomation, pages 8061–8067. IEEE, 2023. 1

    2023

  15. [23]

    pix2gestalt: Amodal segmentation by synthesizing wholes

    Ege Ozguroglu, Ruoshi Liu, D ´ıdac Sur´ıs, Dian Chen, Achal Dave, Pavel Tokmakov, and Carl V ondrick. pix2gestalt: Amodal segmentation by synthesizing wholes. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3931–3940. IEEE Computer Society,

    2024

  16. [24]

    How to make a pizza: Learning a compositional layer-based gan model

    Dim P Papadopoulos, Youssef Tamaazousti, Ferda Ofli, In- gmar Weber, and Antonio Torralba. How to make a pizza: Learning a compositional layer-based gan model. In pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8002–8011, 2019. 2

    2019

  17. [25]

    Instant automatic emptying of panoramic indoor scenes

    Giovanni Pintore, Marco Agus, Eva Almansa, and Enrico Gobbetti. Instant automatic emptying of panoramic indoor scenes. IEEE Transactions on Visualization and Computer Graphics, 28(11):3629–3639, 2022. 1

    2022

  18. [26]

    Learn- ing transferable visual models from natural language super- vision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learn- ing transferable visual models from natural language super- vision. In International Conference on Machine Learning...

    2021

  19. [27]

    Grounded sam: Assembling open-world models for diverse visual tasks

    Tianhe Ren, Shilong Liu, Ailing Zeng, Jing Lin, Kunchang Li, He Cao, Jiayu Chen, Xinyu Huang, Yukang Chen, Feng Yan, et al. Grounded sam: Assembling open-world models for diverse visual tasks. arXiv preprint arXiv:2401.14159 ,

    arXiv

  20. [28]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition, pages 10684–10695, 2022. 5, 6

    2022

  21. [29]

    Laion-400m: Open dataset of clip-filtered 400 million image-text pairs

    Christoph Schuhmann, Richard Vencu, Romain Beaumont, Robert Kaczmarczyk, Clayton Mullis, Aarush Katta, Theo Coombes, Jenia Jitsev, and Aran Komatsuzaki. Laion-400m: Open dataset of clip-filtered 400 million image-text pairs. arXiv preprint arXiv:2111.02114, 2021. 6, 3, 4, 5

    2021 arXiv

  22. [30]

    Zero123++: a single image to consistent multi-view dif- fusion base model

    Ruoxi Shi, Hansheng Chen, Zhuoyang Zhang, Minghua Liu, Chao Xu, Xinyue Wei, Linghao Chen, Chong Zeng, and Hao Su. Zero123++: a single image to consistent multi-view dif- fusion base model. arXiv preprint arXiv:2310.15110, 2023. 8

    arXiv

  23. [31]

    Mulan: A multi layer anno- tated dataset for controllable text-to-image generation

    Petru-Daniel Tudosiu, Yongxin Yang, Shifeng Zhang, Fei Chen, Steven McDonagh, Gerasimos Lampouras, Ignacio Iacobacci, and Sarah Parisot. Mulan: A multi layer anno- tated dataset for controllable text-to-image generation. In Proceedings of the IEEE/CVF Conference on Computer Vi...

    2024

  24. [32]

    Visionllm: Large language model is also an open- ended decoder for vision-centric tasks

    Wenhai Wang, Zhe Chen, Xiaokang Chen, Jiannan Wu, Xizhou Zhu, Gang Zeng, Ping Luo, Tong Lu, Jie Zhou, Yu Qiao, et al. Visionllm: Large language model is also an open- ended decoder for vision-centric tasks. Advances in Neural Information Processing Systems, 36, 2024. 2

    2024

  25. [33]

    Cogvlm: Visual expert for pretrained language models

    Weihan Wang, Qingsong Lv, Wenmeng Yu, Wenyi Hong, Ji Qi, Yan Wang, Junhui Ji, Zhuoyi Yang, Lei Zhao, Xixuan Song, et al. Cogvlm: Visual expert for pretrained language models. Advances in Neural Information Processing Sys- tems, 36, 2024. 2

    2024

  26. [34]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Si- moncelli. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Process- ing, 13(4):600–612, 2004. 6

    2004

  27. [35]

    Amodal com- pletion via progressive mixed context diffusion

    Katherine Xu, Lingzhi Zhang, and Jianbo Shi. Amodal com- pletion via progressive mixed context diffusion. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9099–9109, 2024. 1, 2, 4, 5, 7, 3

    2024

  28. [36]

    Visualizing the invisible: Oc- cluded vehicle segmentation and recovery

    Xiaosheng Yan, Feigege Wang, Wenxi Liu, Yuanlong Yu, Shengfeng He, and Jia Pan. Visualizing the invisible: Oc- cluded vehicle segmentation and recovery. In Proceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 7618–7627, 2019. 2

    2019

  29. [37]

    Amodal ground truth and completion in the wild

    Guanqi Zhan, Chuanxia Zheng, Weidi Xie, and Andrew Zis- serman. Amodal ground truth and completion in the wild. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 28003–28013, 2024. 1 7

    2024

  30. [38]

    Self-supervised scene de- occlusion

    Xiaohang Zhan, Xingang Pan, Bo Dai, Ziwei Liu, Dahua Lin, and Chen Change Loy. Self-supervised scene de- occlusion. In Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition , pages 3784– 3792, 2020. 2

    2020

  31. [39]

    Transparent im- age layer diffusion using latent transparency

    Lvmin Zhang and Maneesh Agrawala. Transparent im- age layer diffusion using latent transparency. ACM Trans. Graph., 43(4), 2024. 1, 2, 8

    2024

  32. [40]

    Face de-occlusion with deep cascade guidance learn- ing

    Ni Zhang, Nian Liu, Junwei Han, Kaiyuan Wan, and Ling Shao. Face de-occlusion with deep cascade guidance learn- ing. IEEE Transactions on Multimedia , 25:3217–3229,

  33. [41]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recogni- tion, pages 586–595, 2018. 6

    2018

  34. [42]

    Visiting the invisible: Layer-by- layer completed scene decomposition

    Chuanxia Zheng, Duy-Son Dao, Guoxian Song, Tat-Jen Cham, and Jianfei Cai. Visiting the invisible: Layer-by- layer completed scene decomposition. International Journal of Computer Vision, 129:3195–3215, 2021. 2

    2021

  35. [43]

    Human de-occlusion: Invisible perception and recovery for humans

    Qiang Zhou, Shiyin Wang, Yitong Wang, Zilong Huang, and Xinggang Wang. Human de-occlusion: Invisible perception and recovery for humans. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 3691–3701, 2021. 2

    2021

  36. [44]

    Semantic amodal segmentation

    Yan Zhu, Yuandong Tian, Dimitris Metaxas, and Piotr Doll´ar. Semantic amodal segmentation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1464–1472, 2017. 5, 6, 3, 4 8

    2017

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.