The DOI matches its Crossref record; 71 of the 88 references it cites were found in a registry or an index; 8 of the 27 sources its claims cite were checked for support.
Vision Transformers for Dense Prediction
Open accessAbstract
We introduce dense prediction transformers, an architecture that leverages vision transformers in place of convolutional networks as a backbone for dense prediction tasks. We assemble tokens from various stages of the vision transformer into image-like representations at various resolutions and progressively combine them into full-resolution predictions using a convolutional decoder. The transformer backbone processes representations at a constant and relatively high resolution and has a global receptive field at every stage. These properties allow the dense prediction transformer to provide finer-grained and more globally coherent predictions when compared to fully-convolutional networks. Our experiments show that this architecture yields substantial improvements on dense prediction tasks, especially when a large amount of training data is available. For monocular depth estimation, we observe an improvement of up to 28% in relative performance when compared to a state-of-the-art fully-convolutional network. When applied to semantic segmentation, dense prediction transformers set a new state of the art on ADE20K with 49.02% mIoU. We further show that the architecture can be fine-tuned on smaller datasets such as NYUv2, KITTI, and Pascal Context where it also sets the new state of the art. Our models are available at https://github.com/intel-isl/DPT.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
32 Own result · 23 Own interpretation · 11 No basis given · 10 No full text of the sources · 3 Supported · 2 Partly supported · 1 Supported so far · 1 Own method · accuracy not yet measured
Every source the paper cites for this claim was checked and supports it.
In the paper
“Specifically, we use the recently proposed vision transformer (ViT) [11] as a backbone architecture.”
§ Vision Transformers for Dense Prediction > 1. Introduction
What the sources say
- Supportedcites [11] 11. Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020. source
Section 3.1 (p. 3) introduces and describes the Vision Transformer (ViT) architecture. The citation identifies the ViT used as the citing authors’ backbone; their choice to use it is not a claim attributed to this source. (LLM, conf=0.93)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Virtually all existing architectures for dense prediction are based on convolutional networks [6,31,34,42,49,50,53].”
§ Vision Transformers for Dense Prediction > 1. Introduction
What the sources say
- Partly supportedcites [6] 6. Liang-Chieh Chen, George Papandreou, Iasonas Kokkinos, Kevin Murphy, and Alan L. Yuille. DeepLab: Semantic im- age segmentation with deep convolutional nets, atrous con- volution, and fully connected crfs. TPAMI, 40(4):834-848, 2018. sourceExperimental — accuracy not yet measured
The Introduction (p. 1) says convolutional models such as PixelCNN had received much more attention for pixel-by-pixel image generation. It does not establish that virtually all architectures across the broader category of dense prediction are convolutional; it also discusses recurrent models and proposes a self-attention alternative (Introduction, p. 2). (LLM, conf=0.78)
- Not checked — no full text of the sourcecites [31] 31. Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U- Net: Convolutional networks for biomedical image segmen- tation. In MICCAI, 2015. source
- Supportedcites [34] 34. Evan Shelhamer, Jonathan Long, and Trevor Darrell. Fully convolutional networks for semantic segmentation. CVPR, 2015. source
The paper is an example of a convolutional-network architecture for dense prediction: its Abstract (p. 1) describes predicting a depth map, and §3.1.2 (p. 3) identifies its convolutional layers and target-depth output. As one of several citations, it supports the sentence by exemplifying such an architecture, rather than establishing the prevalence claim on its own. (LLM, conf=0.88)
- Could not verifycites [42] 42. Jingdong Wang, Ke Sun, Tianheng Cheng, Borui Jiang, Chaorui Deng, Yang Zhao, Dong Liu, Yadong Mu, Mingkui Tan, Xinggang Wang, Wenyu Liu, and Bin Xiao. Deep high-resolution representation learning for visual recogni- tion. TPAMI, 2020. sourceExperimental — accuracy not yet measured
Section 2.1 (p. 8) describes transformer language-model architectures, not architectures for dense prediction. Searches for dense prediction, convolutional networks, and pixel-wise segmentation found no passage addressing whether dense-prediction architectures are predominantly convolutional. (LLM, conf=0.78)
- Not matchedcites [49] 49. Fisher Yu and Vladlen Koltun. Multi-scale context aggrega- tion by dilated convolutions. In ICLR, 2016.
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Not checked — no full text of the sourcecites [50] 50. Yuhui Yuan, Xilin Chen, and Jingdong Wang. Object- contextual representations for semantic segmentation. In ECCV, 2020. source
- Not checked — no full text of the sourcecites [53] 53. Hengshuang Zhao, Jianping Shi, Xiaojuan Qi, Xiaogang Wang, and Jiaya Jia. Pyramid scene parsing network. In CVPR, 2017. source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“We notice that the biggest gains in performance for zero-shot transfer were achieved for datasets that feature dense, high-resolution evaluations [15,55,59].”
§ Vision Transformers for Dense Prediction > Additional information
What the sources say
- Not checked — no full text of the sourcecites [15] 15. Andreas Geiger, Philip Lenz, and Raquel Urtasun. Are we ready for autonomous driving? The KITTI vision benchmark suite. In CVPR, 2012. source
- Supportedcites [55] 55. D. J. Butler, J. Wulff, G. B. Stanley, and M. J. Black. A naturalistic open source movie for optical flow evaluation. In ECCV, 2012. source
The citation identifies MPI-Sintel as one of the datasets in the citing authors’ comparison; the performance gains are the citing authors’ observation, not a result attributed to this paper. Design Decisions and Comparison to Middlebury (p. 5) describes Sintel’s higher-resolution frames, and The Sintel Data Set (p. 9) describes its flow fields and pixel-level error evaluation. (LLM, conf=0.86)
- Not checked — no full text of the sourcecites [59] 59. Thomas Schöps, Johannes L. Schönberger, Silvano Galliani, Torsten Sattler, Konrad Schindler, Marc Pollefeys, and An- dreas Geiger. A multi-view stereo benchmark with high- resolution images and multi-camera videos. In CVPR, 2017. source
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Note that ResNext101-WSL was pretrained on a billion-scale corpus of weakly supervised data [25] in addition to ImageNet pretraining.”
§ Vision Transformers for Dense Prediction > Skip connections.
What the sources say
- Partly supportedcites [25] 25. Dhruv Mahajan, Ross Girshick, Vignesh Ramanathan, Kaiming He, Manohar Paluri, Yixuan Li, Ashwin Bharambe, and Laurens van der Maaten. Exploring the limits of weakly supervised pretraining. In ECCV, 2018. sourceExperimental — accuracy not yet measured
Section 2.3 (p. 4) identifies the models as ResNeXt-101, and Section A.5 (p. 23) says they were pretrained on billions of Instagram images using hashtags as labels. But it describes subsequent training on ImageNet as fine-tuning, not additional ImageNet pretraining; Section 3.2 (p. 12) also distinguishes Instagram-pretrained models from ImageNet-pretrained baselines. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“While ignoring the token yields good performance, projection provides slightly better performance on average.”
§ Vision Transformers for Dense Prediction > Skip connections.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Adding the token, on the other hand, yields worse performance than simply ignoring it.”
§ Vision Transformers for Dense Prediction > Skip connections.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Both DPT variants significantly outperform the state of the art.”
§ Vision Transformers for Dense Prediction > 4.1. Monocular Depth Estimation
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Note that both architectures have similar latency to MiDaS (Table 9).”
§ Vision Transformers for Dense Prediction > 4.1. Monocular Depth Estimation
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“ViT-Large outperforms all other backbones but is also almost three times larger than ViT-Base and ViT-Hybrid.”
§ Vision Transformers for Dense Prediction > Skip connections.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We again see that DPT can provide strong performance even on smaller datasets.”
§ Vision Transformers for Dense Prediction > ADE20K.
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Probably not a substantial claim. 11
- It may state two separate claims as one. 13
- The quote only partly states this claim. 1
Every source the paper cites for this claim was checked and supports it.
In the paper
“In particular, it has recently been demonstrated that a direct application of token-based transformer architectures that have been successful in NLP can yield competitive performance on image classification [11].”
§ Vision Transformers for Dense Prediction > 2. Related Work
What the sources say
- Supportedcites [11] 11. Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020. source
The Introduction (p. 1) says image patches are treated like NLP tokens and fed to a standard Transformer. The Conclusion (p. 9) reports that this approach matches or exceeds the state of the art on many image-classification datasets when coupled with pre-training on large datasets. (LLM, conf=0.93)
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“A key insight of this work was that, like transformer models in NLP, vision transformers need to be paired with a sufficient amount of training data to realize their potential.”
§ Vision Transformers for Dense Prediction > 2. Related Work
What the sources say
- Supportedcites [11] 11. Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020. source
The Introduction (p. 1) says ViT performs modestly on mid-sized datasets and does not generalize well with insufficient data, but attains excellent results when pre-trained at sufficient scale. Section 4.3 (p. 6) reports that larger ViT models show their full benefit only with the largest pre-training dataset tested. The Conclusion (p. 9) also describes the approach as working well when coupled with pre-training on large datasets. (LLM, conf=0.94)
plainconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
- Supported
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Similarly, the ViT-Hybrid architecture extracts features at 1 16 the input resolution, which is twice as high as the lowest-resolution features that are commonly used with convolutional backbones.”
§ Vision Transformers for Dense Prediction > 3. Architecture
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“It contains about 1.4 million images and is, to the best of our knowledge, the largest training set for monocular depth estimation that has ever been compiled.”
§ Vision Transformers for Dense Prediction > 4.1. Monocular Depth Estimation
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“When applied to semantic segmentation, dense vision transformers set a new state of the art on ADE20K with 49.02% mIoU.”
§ Vision Transformers for Dense Prediction > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The quote only partly states this claim.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We observe that the DPT tends to produce cleaner and finer-grained delineations of object boundaries and that the predictions are also in some cases less cluttered.”
§ Vision Transformers for Dense Prediction > ADE20K.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“DPT can better reconstruct fine details while also improving global coherence in areas that are challenging for the convolutional architecture (for example, large homogeneous regions or relative depth arrangement across the image).”
§ Vision Transformers for Dense Prediction > 4.1. Monocular Depth Estimation
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The average relative improvement over the best published architecture, MiDaS, is more than 23% for DPT-Hybrid and 28% for DPT-Large.”
§ Vision Transformers for Dense Prediction > 4.1. Monocular Depth Estimation
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“DPT-Hybrid achieves this with a comparable network capacity (Table 9), while DPT-Large is about 3 times larger than MiDaS.”
§ Vision Transformers for Dense Prediction > 4.1. Monocular Depth Estimation
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We observe the tendency that attention is spatially more localized close to the reference token in shallow layers (leftmost columns), whereas deeper layers (rightmost columns) frequently attend across the whole image.”
§ Vision Transformers for Dense Prediction > Additional information
hedgedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“ViT-Base has comparable performance to ResNext101-WSL, while ViT-Hybrid and ViT-Large improve performance even though they have been pretrained on significantly less data.”
§ Vision Transformers for Dense Prediction > Skip connections.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“While we observe a general trend of an improvement in per-class IoU in comparison to the baseline [51], we do not observe a strong pattern across classes.”
§ Vision Transformers for Dense Prediction > Additional information
What the sources say
- Not checked — no full text of the sourcecites [51] 51. Hang Zhang, Chongruo Wu, Zhongyue Zhang, Yi Zhu, Zhi Zhang, Haibin Lin, Yue Sun, Tong He, Jonas Muller, R. Manmatha, Mu Li, and Alexander Smola. ResNeSt: Split- attention networks. arXiv preprint arXiv:2004.08955, 2020.
hedgedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“While we observe a general trend of an improvement in per-class IoU in comparison to the baseline [51], we do not observe a strong pattern across classes.”
§ Vision Transformers for Dense Prediction > Additional information
What the sources say
- Not checked — no full text of the sourcecites [51] 51. Hang Zhang, Chongruo Wu, Zhongyue Zhang, Yi Zhu, Zhi Zhang, Haibin Lin, Yue Sun, Tong He, Jonas Muller, R. Manmatha, Mu Li, and Alexander Smola. ResNeSt: Split- attention networks. arXiv preprint arXiv:2004.08955, 2020.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We observe more details and also better global depth arrangement in DPT predictions when compared to the fullyconvolutional baseline.”
§ Vision Transformers for Dense Prediction > Additional information
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The transformer backbone processes representations at a constant and relatively high resolution and has a global receptive field at every stage.”
§ Vision Transformers for Dense Prediction > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“As long as the image size is divisible by p, the embedding procedure can be applied and will produce a varying number of image tokens N p .”
§ Vision Transformers for Dense Prediction > 3. Architecture
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Unlike fully-convolutional networks, the vision transformer backbone foregoes explicit downsampling operations after an initial image embedding has been computed and maintains a representation with constant dimensionality throughout all processing stages.”
§ Vision Transformers for Dense Prediction > 1. Introduction
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Since transformers are known to realize their full potential only when an abundance of training data is available, monocular depth estimation is an ideal task to test the capabilities of DPT.”
§ Vision Transformers for Dense Prediction > 4.1. Monocular Depth Estimation
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“This is in stark contrast to convolutional networks, which progressively increase their receptive field as features pass through consecutive convolution and downsampling layers.”
§ Vision Transformers for Dense Prediction > 3. Architecture
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“DPT-Large performs slightly worse, likely because of the significantly smaller dataset compared to our previous experiments.”
§ Vision Transformers for Dense Prediction > ADE20K.
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“DeIT trains the ViT architecture with a more data-efficient pretraining procedure.”
§ Vision Transformers for Dense Prediction > Skip connections.
What the sources say
- Not checked — no full text of the sourcecites [38] 38. Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, and Hervé Jégou. Training data-efficient image transformers & distillation through at- tention. arXiv preprint arXiv:2012.12877, 2020. source
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Note that the DeIT-Base architecture is identical to ViT-Base, while DeIT-Base-Dist introduces an additional distillation token, which we ignore in the Reassemble operation.”
§ Vision Transformers for Dense Prediction > Skip connections.
What the sources say
- Not checked — no full text of the sourcecites [38] 38. Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, and Hervé Jégou. Training data-efficient image transformers & distillation through at- tention. arXiv preprint arXiv:2012.12877, 2020. source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Similar to prior work on transformers, DPT unfolds its full potential when trained on large-scale datasets.”
§ Vision Transformers for Dense Prediction > 5. Conclusion
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“Note that results for DPT and Mi-DaS are computed at the same input resolution (384 pixels).”
§ Vision Transformers for Dense Prediction > Additional information
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
The paper relies on this without support.
The choice of backbone architecture has a large influence on the capabilities of the overall model, because information lost in the encoder cannot be recovered in the decoder.
“However, it is widely recognized that the choice of backbone architecture has a large influence on the capabilities of the overall model, as any information that is lost in the encoder is impossible to recover in the decoder.”
The paper relies on this without support.
Feature resolution and granularity are critical for dense prediction, where the architecture should ideally resolve features at or close to the input resolution.
“While feature resolution and granularity may not matter for some tasks, such as image classification, they are critical for dense prediction, where the architecture should ideally be able to resolve features at or close to the resolution of the input image.”
The paper relies on this without support.
Convolutions are linear operators with a limited receptive field.
“Convolutions, by definition, are linear operators that have a limited receptive field.”
The paper relies on this without support.
Downsampling intermediate representations is necessary to keep memory consumption at levels feasible with existing computer architectures.
“Downsampling the intermediate representations is necessary to keep memory consumption at levels that are feasible with existing computer architectures.”
The paper relies on this without support.
The transformer backbone maintains a constant feature resolution.
“Since the transformer backbone maintains a constant feature resolution, it is not clear at which points in the backbone features should be tapped.”
The paper relies on this without support.
Performance of fully-convolutional architectures suffers heavily when performing inference at an input resolution significantly different from the training resolution.
“Performance thus suffers heavily when performing inference at an input resolution that is significantly different from the training resolution.”
The paper relies on this without support.
Transformer encoders have a global receptive field in every layer.
“Transformer encoders, on the other hand, have a global receptive field in every layer.”
The paper relies on this without support.
The reduced meta-dataset datasets provide high-quality ground truth.
“We choose these datasets since they provide high-quality ground truth.”
The paper relies on this without support.
Convolutional architectures offer natural points of interest for passing features from the encoder to the decoder, namely before or after downsampling of the representation.
“Convolutional architectures offer natural points of interest for passing features from the encoder to the decoder, namely before or after downsampling of the representation.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Vision transformers, which are a deep learning approach to encoding low-dimensional input into a high dimensional feature space, have established new frontiers in depth estimation compared to convolutional neural networks (Ranftl et al., 2021).”
From Very high resolution canopy height maps from RGB imagery using self-supervised vision transformer and convolutional decoder trained on aerial lidar (2024) · cites this paper as Ranftl et al., 2021
In the citing paper
“However, the Transformer framework does not involve dimensionality reduction10,12, thus keeping all attention layer traceable to the original input features13, thus making the models interpretable.”
From Transformer for one stop interpretable cell type annotation (2023) · cites this paper as [10]
In the citing paper
“For example, the change of dimensions and non-linear aggregation of features throughout the autoencoder’s deep processing stages leads to untraceable and uninterpretable latent space and loss of information and feature resolution9,10.”
From Transformer for one stop interpretable cell type annotation (2023) · cites this paper as [10]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Vision Transformers for Dense Prediction > 4.1. Monocular Depth Estimation
“Our architecture matches or improves state-of-the-art performance on both datasets in all metrics.”
§ Vision Transformers for Dense Prediction > 5. Conclusion
“| BTS [20] | 0.956 | 0.993 | 0.998 | 0.059 2.756 | 0.096 | | DPT-Hybrid 0.959 | 0.995 | 0.999 | 0.062 2.573 | 0.092 |”
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 4×
“We perform a similar experiment with the hybrid architecture in Table 6 (bottom), where R0 and R1 refer to us-ing features from the first and second downsampling stages of the ResNet50 embedding network.”
MethodInfluentialVaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
U-Net: Convolutional Networks for Biomedical Image Segmentation
Ronneberger, O., Fischer, P., & Brox, T. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-24574-4_28 · Cited by 99,404 · cited 3×
“These include training at higher input resolution (if the computational budget permits), dilated convolutions [51] to rapidly increase the receptive field without downsampling, appropriately-placed skip connections from multiple stages of the encoder to the decoder [33], or, more recently, by connecting multiresolutio…”
BackgroundAdam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
ImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333 · cited 5×
“The encoder is frequently based on an image classification network, also called the backbone, that is pretrained on a large corpus such as ImageNet [9].”
InfluentialBERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325 · cited 6×
“Following work in NLP, the ViT additionally adds a special token that is not grounded in the input image and serves as the final, global image representation which is used for classification.”
BackgroundMethodInfluentialAn Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 47,472 · cited 8×
“On a high level, the vision transformer (ViT) [11] operates on a bag-of-words representation of the image [38].”
BackgroundMethodResultInfluentialFully convolutional networks for semantic segmentation
Long, J., Shelhamer, E., & Darrell, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298965 · Cited by 38,834 · cited 2×
“Virtually all existing architectures for dense prediction are based on convolutional networks [6, 33, 36, 44, 51, 52, 55].”
BackgroundLanguage Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Research works printed but not linked: 3
- yuan, 2020, Objectcontextual representations for semantic segmentation, ECCV
- godard, 2017, Unsupervised monocular depth estimation with leftright consistency, CVPR
- casser, 2019, Unsupervised learning of depth and ego-motion: A structured approach, AAAI
Cited by
Sorted by citations
Items 301–400 of 2,248
AdaptorNAS: A New Perturbation-Based Neural Architecture Search for Hyperspectral Image Segmentation
Ang, S. P., Phung, S. L., Bui, L., & Bouzerdoum, A. (2023) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2023.3298796 · Cited by 27
“In this section, AdaptorNAS is compared to five state-of-the-art deep learning methods of diverse methodologies : 1) DPT [63]: DPT converts the tokens from different stages of the vision transformer (ViT) into image-like representations of various resolutions.”
MethodGasMono: Geometry-Aided Self-Supervised Monocular Depth Estimation for Indoor Scenes
Zhao, C., Poggi, M., Tosi, F., Zhou, L., Sun, Q., Tang, Y., & Mattoccia, S. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01485 · Cited by 35
“Among many, the possibility of learning to estimate depth out of a single image – for long considered the holy grail in computer vision – be-∗ came true [7, 9, 23, 38, 39].”
BackgroundFine-Tuning Image-Conditional Diffusion Models is Easier than you Think
Garcia, G. M., Zeid, K. A., Schmidt, C., de Geus, D., Hermans, A., & Leibe, B. (2025) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61041.2025.00083 · Cited by 61
“When training Marigold for normal prediction as a diffusion estimator, the mask is downsam-pled by a factor of 8 to match the latent resolution.”
A data-driven method for online transient stability monitoring with vision-transformer networks
Fang, J., Liu, C., Zheng, L., & Su, C. (2023) · International Journal of Electrical Power & Energy Systems · doi:10.1016/j.ijepes.2023.109020 · Cited by 32
Shen, L., Zhang, L., Qi, P., et al. (2025) · PhotoniX · doi:10.1186/s43074-025-00185-4 · Cited by 26
“We compare our proposed method with three methods, including two binocular image depth estimation methods: disparitySGM [34], UniMatch [38], and one monocular image depth estimation method: Dpt-large [39].”
Cerberus Transformer: Joint Semantic, Affordance and Attribute Parsing
Chen, X., Liu, T., Zhao, H., Zhou, G., & Zhang, Y.-Q. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01903 · Cited by 28
“Following [22], an extra learnable token t0 is added to the sequence, which serves for attention visualization in Fig.”
BackgroundMethodMultimodal foundation model and benchmark for comprehensive retinal OCT image analysis
Morano, J., Fazekas, B., Sükei, E., Fecso, R., Emre, T., Gumpinger, M., Faustmann, G., Oghbaie, M., Schmidt-Erfurth, U. M., & Bogunović, H. (2025) · npj Digital Medicine · doi:10.1038/s41746-025-01852-3 · Cited by 39
Booster: A Benchmark for Depth From Images of Specular and Transparent Surfaces
Ramirez, P. Z., Costanzino, A., Tosi, F., Poggi, M., Salti, S., Mattoccia, S., & Stefano, L. D. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3323858 · Cited by 38
“However, if the base network presents large diffuse errors, such as in the third Finetuning by the Booster Training Data: As we did for stereo networks, we select two competitive methods (here MiDaS [13] and DPT [2]) for fine-tuning them on the Booster Stereo training set.”
BackgroundMethodInfluentialGuan, T., He, Z., Song, R., Manocha, D., & Zhang, L. (2023) · Autonomous Robots · doi:10.1007/s10514-023-10113-9 · Cited by 26
TMVNet : Using Transformers for Multi-view Voxel-based 3D Reconstruction
Peng, K., Islam, R., Quarles, J. P., & Desai, K. (2022) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw56347.2022.00036 · Cited by 25
“Another benefit is that the transformer could also learn object’s 3D position feature from image [21].”
BackgroundLuginov, A., & Makarov, I. (2023) · IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-Adjunct) · doi:10.1109/ismar-adjunct60411.2023.00137 · Cited by 28
“Ranftl et al. [36] were the first to introduce ViT [6] in supervised MDE, outperforming CNN-based solutions.”
MethodVision Transformers for Vein Biometric Recognition
Garcia-Martin, R., & Sánchez-Reillo, R. (2023) · IEEE Access · doi:10.1109/access.2023.3252009 · Cited by 27
“Additionally, they can be found in more challenging missions like image segmentation (semantic, instance, and panoptic) [14], [15], [16] or even image generation with the introduction of ViTGAN [17].”
BackgroundDepth Inpainting via Vision Transformer
Makarov, I., & Borisenko, G. V. (2021) · IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-Adjunct) · doi:10.1109/ismar-adjunct54149.2021.00065 · Cited by 27
Delving into Shape-aware Zero-shot Semantic Segmentation
Liu, X., Tian, B., Wang, Z., Wang, R., Sheng, K., Zhang, B., Zhao, H., & Zhou, G. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00293 · Cited by 27
“Visual Encoder We employ Dilated residual networks (DRN) [57] and Dense Prediction Transformers (DPT) [43] to encoder images into pixel-level embeddings.”
MethodInfluentialMulti-scale and Cross-scale Contrastive Learning for Semantic Segmentation
Pissas, T., Ravasio, C. S., da Cruz, L., & Bergeles, C. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19818-2_24 · Cited by 28
Ferracani, A., Bertini, M., Pala, P., Nannotti, G., Principi, F., & Becchi, G. (2024) · doi:10.1145/3689094.3689465 · Cited by 31
MVPGS: Excavating Multi-view Priors for Gaussian Splatting from Sparse Input Views
Xu, W., Gao, H., Shen, S., Peng, R., Jiao, J., & Wang, R. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72970-6_12 · Cited by 25
Learning Stereo Depth Estimation with Bio-Inspired Spike Cameras
Wang, Y., Li, J., Zhu, L., Xiang, X., Huang, T., & Tian, Y. (2022) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme52920.2022.9859975 · Cited by 28
“The emerging transformers [12, 13] revisit the problem from a sequence-to20 22 IE EE In te rn at io na l C on fe re nc e on M ul tim ed ia a nd E xp o (IC M E) | 9 78 -1 -6 65 485 63 -0 /2 2/ $3 1.”
BackgroundDevulapally, A., Khan, M. F. F., Advani, S., & Narayanan, V. (2024) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw63382.2024.00213 · Cited by 29
“Unlike the conventional meth-ods [18] which uses multiple transformer blocks in the encoder phase, we use single transformer and perform the patch embedding on the feature space generated from con-vLSTM outputs rather than the raw events.”
MethodRT-ViT: Real-Time Monocular Depth Estimation Using Lightweight Vision Transformers
Ibrahem, H., Salem, A., & Kang, H. (2022) · Sensors · doi:10.3390/s22103849 · Cited by 25
“DPT [25] adopts a different encoder since the authors used the large ViT (ViT-L) with 24 transformer layers however in our implementation we use far simpler transformers with 12, 6, and 4 transformer layers as we proved that the excess use of transformer layers is redundant in some cases where the ViT-s16 and ViT-t16…”
MethodInfluentialTowards Comprehensive Monocular Depth Estimation: Multiple Heads are Better Than One
Shao, S., Li, R., Pei, Z., Liu, Z., Chen, W., Zhu, W., Wu, X., & Zhang, B. (2022) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2022.3224810 · Cited by 25
“In contrast to the prior ensemble methods, we explore the ensembles of visual transformers and utilize the complementarity of CNN and Transformer to enforce the diversity among base predictors.”
MethodLow-Resolution Self-Attention for Semantic Segmentation
Wu, Y.-H., Zhang, S., Liu, Y., Zhang, L., Zhan, X., Zhou, D., Feng, J., Cheng, M., & Zhen, L. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3577035 · Cited by 36
Enhancing furcation involvement classification on panoramic radiographs with vision transformers
Zhang, X., Guo, E., Liu, X., Zhao, H., Yang, J., Li, W., Wu, W.-L., & Sun, W.-B. (2025) · BMC Oral Health · doi:10.1186/s12903-025-05431-6 · Cited by 25
“The ViT represents a novel approach to image classification by leveraging the transformer architecture, which was originally developed for natural language processing [26].”
PCTNet: depth estimation from single structured light image with a parallel CNN-transformer network
Zhu, X., Han, Z., Zhang, Z., Song, L., Wang, H., & Guo, Q. (2023) · Measurement Science and Technology · doi:10.1088/1361-6501/acd136 · Cited by 24
Fedhca 2 : Towards Hetero-Client Federated Multi-Task Learning
Lu, Y., Huang, S., Yang, Y., Sirejiding, S., Ding, Y., & Lu, H. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.00535 · Cited by 25
“While existing FMTL approaches primarily address statistical challenges [51, 65], recent studies [8, 12, 92] have highlighted the importance of task heterogeneity , particularly for dense predictions such as semantic segmentation and depth estimation [11, 58, 74].”
BackgroundCSDFormer: A cloud and shadow detection method for landsat images based on transformer
Li, J., & Wang, Q. (2024) · International Journal of Applied Earth Observation and Geoinformation · doi:10.1016/j.jag.2024.103799 · Cited by 26
TwinsTNet: Broad-View Twins Transformer Network for Bi-Modal Salient Object Detection
Lyu, P., Yu, X., Chi, J., Wu, H., Wu, C., & Rajapakse, J. C. (2025) · IEEE Transactions on Image Processing · doi:10.1109/tip.2025.3564821 · Cited by 28
“In contrast, the Trans-former [25] has recently emerged as a promising alternative, outperforming CNNs in several mainstream computer vision tasks [26], [27], [28], [29].”
IINet: Implicit Intra-inter Information Fusion for Real-Time Stereo Matching
Li, X., Zhang, C., Su, W., et al. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i4.28107 · Cited by 24
WorDepth: Variational Language Prior for Monocular Depth Estimation
Zeng, Z., Wang, D., Yang, F., Park, H., Soatto, S., Lao, D., & Wong, A. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.00927 · Cited by 36
“Monocular depth estimation trains by minimizing loss between depth predictions and ground-truth depth maps [2, 7, 17, 35, 46, 52, 54, 61, 66, 78, 80, 84, 86].”
BackgroundExploring Sparse Visual Prompt for Domain Adaptive Dense Prediction
Yang, S., Wu, J., Liu, J., Li, X., Zhang, Q., Pan, M., Gan, Y., Chen, Z., & Zhang, S. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i15.29569 · Cited by 39
“Overall, our method outperforms several previous SOTA methods on all semantic segmentation TTA and CTTA tasks and shows promising potential for real-world applications.”
MethodAttention combined pyramid vision transformer for polyp segmentation
Liu, X., & Song, S. (2023) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2023.105792 · Cited by 24
SegLoc: Learning Segmentation-Based Representations for Privacy-Preserving Visual Localization
Pietrantoni, M., Humenberger, M., Sattler, T., & Csurka, G. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.01476 · Cited by 25
“In the initialization step, reference images are processed by the initial SegLoc network (where the parameters are initialized with the pre-trained DPT) and dense features from the encoder are sampled and their associated ADE20k predictions are collected.”
MethodInfluentialWang, Y., Pan, Z., Li, X., Cao, Z., Xian, K., & Zhang, J. (2022) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3503161.3547978 · Cited by 33
“In recent years, the spatial accuracy of single image depth estimation [28, 41, 42, 54] has been significantly improved.”
BackgroundMaterialistic: Selecting Similar Materials in Images
Sharma, P., Philip, J., Gharbi, M., Freeman, B., Durand, F., & Deschaintre, V. (2023) · ACM Transactions on Graphics · doi:10.1145/3592390 · Cited by 28
“…been used in several vision tasks, such as image classification [Chen et al. 2020; Dosovitskiy et al. 2020; Hu et al. 2018], semantic segmentation [Ranftl et al. 2021; Wang et al. 2021, 2020b], super-resolution [Cao et al. 2021; Chen et al. 2022; Lu et al. 2021; Yang et al. 2020], image…”
MethodSingle-view 3D Scene Reconstruction with High-fidelity Shape and Texture
Chen, Y., Ni, J., Jiang, N., Zhang, Y., Zhu, Y., & Huang, S. (2024) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv62453.2024.00142 · Cited by 24
“[62].”
Bavirisetti, D. P., Martinsen, H. R., Kiss, G. H., & Lindseth, F. (2023) · IEEE Open Journal of Intelligent Transportation Systems · doi:10.1109/ojits.2023.3335648 · Cited by 28
“InMarch 2021, Ranftl et al. proposed the Dense Prediction Transformer (DPT) [31], a novel architecture for the tasks of monocular depth estimation and semantic segmentation.”
BackgroundA Study on the Generality of Neural Network Structures for Monocular Depth Estimation
Bae, J., Hwang, K., & Im, S. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3332407 · Cited by 27
“Recently, Transformers [32] has shown promise for solving computer vision tasks such as image classification [60], [61], object detection [33], and dense prediction [11], [62], [63], [64].”
BackgroundMethodInfluentialG2-MonoDepth: A General Framework of Generalized Depth Inference From Monocular RGB+X Data
Wang, H., Yang, M., & Zheng, N. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3346466 · Cited by 23
“Notably, we adopt a basic U-Net [60] with low complexity to establish a basic benchmark for our unified task, rather than other complicated architectures such as DenseNet [49], HourglassNet [7], or ViTs [44] [45].”
MethodOnline Knowledge Distillation for Multi-task Learning
Jacob, G. M., Agarwal, V., & Stenger, B. (2023) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv56688.2023.00239 · Cited by 24
“Vision transformers (ViT) have emerged as a successful technique for many tasks, such as image classification [25, 26], object detection [3, 27] and pixel-wise prediction problems such as depth estimation [28, 29] and semantic segmentation [30].”
MethodChamola, V., Chougule, A., Sam, A., Hussain, A., & Yu, F. R. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3362851 · Cited by 26
“The performance of several models, such as BinsFormer [192], Depthformer [193], MonoDELSNet [194], SfM-Revisited [195], AdaBins [196], DORN [197], DPT-Hybrid [198], GCNDepth [199] SC-Depth [200], and DNET [201], that employed the KITTI data set for depth estimation is shown in Fig.”
MethodPyramidal Attention for Saliency Detection
Hussain, T., Anwar, A., Anwar, S., et al. (2022) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw56347.2022.00325 · Cited by 23
Robust and Efficient Modulation Recognition with Pyramid Signal Transformer
He, S., Fan, X., & Liu, H. (2022) · Globecom · doi:10.1109/globecom48099.2022.10001593 · Cited by 23
“Self-attention [9] and Transformer [10–12] as novel in-ventions in natural language processing (NLP) and computer vision (CV) communities have shown great potential to classify the radio signals [13].”
BackgroundGLPanoDepth: Global-to-Local Panoramic Depth Estimation
Bai, J.-C., Qin, H., Lai, S.-C., Guo, J., & Guo, Y.-W. (2022) · IEEE Transactions on Image Processing · doi:10.1109/tip.2024.3386403 · Cited by 41
“[33] also tried to recover dense depth values from a single image by vision transformers.”
BackgroundDepth Map Decomposition for Monocular Depth Estimation
Jun, J., Lee, J.-H., Lee, C., & Kim, C. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-20086-1_2 · Cited by 29
When Multitask Learning Meets Partial Supervision: A Computer Vision Review
Fontana, M., Spratling, M., & Shi, M. (2024) · Proceedings of the IEEE · doi:10.1109/jproc.2024.3435012 · Cited by 27
“First, the depth modality is approximated by running a pretrained DPT-Hybrid [181], a ViT-based model.”
MethodA Survey on Self-Supervised Monocular Depth Estimation Based on Deep Neural Networks
Dong, Q., Zhou, Z., Qiu, X., & Zhang, L. (2025) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2025.3552598 · Cited by 29
“Inrecent years, with the development of deep learning techniques and large-scale datasets, the monocular depth estimation methods based on deep neural networks (DNNs) have achieved remarkable performances [12], [13], [14], [15], [16].”
SPDET: Edge-Aware Self-Supervised Panoramic Depth Estimation Transformer With Spherical Geometry
Zhuang, C., Lu, Z., Wang, Y., Xiao, J., & Wang, Y. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3272949 · Cited by 25
“ViT [36] and DPT [37] divide an image into patches and embed patches to tokens, then”
BackgroundMethodInfluentialSelf-Supervised Depth Estimation Leveraging Global Perception and Geometric Smoothness
Jia, S., Pei, X., Yao, W., & Wong, S. C. (2022) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2022.3219604 · Cited by 27
“Numerous recent studies have applied Transformer to computer vision tasks by reshaping square images to sequencelike data, either using Transformer for CNN feature processing [52], [53], [54], [55], [56] or for feature extraction [57], [58], [59], [60], [61].”
MethodSun, Y., Liu, B., Yu, X., Yu, A., Xue, Z., & Gao, K. (2022) · International Journal of Remote Sensing · doi:10.1080/01431161.2022.2054299 · Cited by 24
“While multiple resolutions of feature and granularity information are generally crucial for dense prediction tasks (Ranftl, Bochkovskiy and Koltun 2021).”
BackgroundRen, Y., Wang, L., Li, M., Jiang, H., Cui, Z., Yang, M., Yu, H., & Yang, D. (2025) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2025.3606554 · Cited by 29
Scene‐Aware 3D Multi‐Human Motion Capture from a Single Camera
Luvizon, D. C., Habermann, M., Golyanik, V., Kortylewski, A., & Theobalt, C. (2023) · Computer Graphics Forum · doi:10.1111/cgf.14768 · Cited by 24
“First, we obtain normalized disparity maps d̂t from the state-of-the-art DPT model [41], which are then postprocessed to enhance sharpness [58].”
BackgroundMethodDeep Learning for Omnidirectional Vision: A Survey and New PerspectivesPreprint
Ai, H., Cao, Z., Zhu, J., Bai, H., Chen, Y., & Wang, L. (2022) · doi:10.36227/techrxiv.19807699 · Cited by 36
Gonçalves, D. N., Marcato, J., Zamboni, P., et al. (2023) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw59228.2023.00669 · Cited by 22
“Since ViT, several transformer-based networks with prominent results have been proposed for image classification [5,8,36], object detection [3, 43] and image segmentation [27, 31, 35, 40].”
BackgroundNTIRE 2024 Challenge on HR Depth from Images of Specular and Transparent Surfaces
Ramirez, P. Z., Tosi, F., Stefano, L. D., Timofte, R., Costanzino, A., Poggi, M., Salti, S., Mattoccia, S., Zhang, Y., Wu, C., et al. (2024) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw63382.2024.00650 · Cited by 21
“MiDaS [60] represents the pivotal work in this direction, training a CNN on a mixture of several datasets to achieve cross-domain generalization – followed by DPT [58], Depth Anything [93], and Marigold [28].”
BackgroundTransformer-Based Self-Supervised Monocular Depth and Visual Odometry
Zhao, H., Qiao, X., Ma, Y., & Tafazolli, R. (2022) · IEEE Sensors Journal · doi:10.1109/jsen.2022.3227017 · Cited by 22
“DPT [23] leverages pure transformer as the basic computational block of the encoder, while TransDepth [25] uses a hybrid encoder.”
BackgroundMethodLi, J., Wang, J., & Niu, J. (2023) · Building and Environment · doi:10.1016/j.buildenv.2023.110912 · Cited by 22
MoD-SLAM: Monocular Dense Mapping for Unbounded 3D Scene Reconstruction
Zhou, H., Guo, Z., Ren, Y., Liu, S., Zhang, L., Zhang, K., & Li, M. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3498777 · Cited by 27
“In the relative depth estimation module, we adopt the training criterion of MiDaS [24] and use the DPT decoder-encoder architecture [23] as backbones for feature extraction to predict the relative depths.”
MethodUnified Transformer Network for Multi-Weather Image Restoration
Kulkarni, A., Phutke, S. S., & Murala, S. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-25063-7_21 · Cited by 25
Opportunities and Challenges of Foundation Models in Industrial Manufacturing
Ayyat, M., Osman, M., & Nadeem, T. (2025) · IEEE Access · doi:10.1109/access.2025.3589777 · Cited by 25
“Examples include state-of-the-art results in depth estimation [84], [85], ImageNet classification [86], and image segmentation [38], among many others.”
Diffusion Models for Monocular Depth Estimation: Overcoming Challenging Conditions
Tosi, F., Ramirez, P. Z., & Poggi, M. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73337-6_14 · Cited by 34
Tschannen, M., Gritsenko, A., Wang, X., Naeem, M. F., Alabdulmohsin, I., Parthasarathy, N., Evans, T., Beyer, L., Xia, Y., Mustafa, B., et al. (2025) · arXiv · doi:10.48550/arxiv.2502.14786 · Cited by 22
WRRT-DETR: Weather-Robust RT-DETR for Drone-View Object Detection in Adverse Weather
Liu, B., Jin, J., Zhang, Y., & Sun, C. (2025) · Drones · doi:10.3390/drones9050369 · Cited by 21
“Since the PureSea maritime dataset lacks available depth maps, we utilized the pre-trained depth estimation model DPT [43] to predict depth information D ( x , y ) .”
PSDR-Room: Single Photo to Scene using Differentiable Rendering
Yan, K., Luan, F., Hašan, M., et al. (2023) · SIGGRAPH Asia Conference Papers · doi:10.1145/3610548.3618165 · Cited by 21
“We utilize models for intrinsic camera parameter [Lopez et al. 2019; Zhang et al. 2020b; Jin et al. 2023] and depth estimation [Ranftl et al. 2020, 2021].”
MethodInfluentialKhan, F., Hussain, S., Basak, S., Lemley, J., & Corcoran, P. (2021) · Neural Networks · doi:10.1016/j.neunet.2021.07.007 · Cited by 21
Flare-Free Vision: Empowering Uformer with Depth Insights
Kotp, Y., & Torki, M. (2024) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp48485.2024.10446006 · Cited by 21
“As shown in Figure 3, our approach consists of two stages, first is a depth estimation where the depth map for the image is estimated using a dense vision transformer (DPT) [5, 6] DPT uses an encoder-decoder structure which is the most successful structure in dense prediction.”
MethodBodyMap: Learning Full-Body Dense Correspondence Map
Ianina, A., Sarafianos, N., Xu, Y., Rocco, I., & Tung, T. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01293 · Cited by 22
“[34] for monocular depth estimation and introduce a simple yet novel transformer-based architecture designed explicitly for the task of predicting dense surface correspondences of humans.”
BackgroundMethodSemantic Segmentation of UAV Images Based on Transformer Framework with Context Information
Kumar, S., Kumar, A., & Lee, D.-G. (2022) · Mathematics · doi:10.3390/math10244735 · Cited by 24
“PVT [27], DPT [43], and Segformer [44] adopted different design choices to generate multi-scale feature representations like CNNs.”
BackgroundDeep transformer networks for precise pothole segmentation tasks
Katsamenis, I., Sakelliou, A., Bakalos, N., et al. (2023) · International Conference on PErvasive Technologies Related to Assistive Environments (PETRA) · doi:10.1145/3594806.3596560 · Cited by 21
“This segmentation model is based on a ground-breaking backbone architecture for vision tasks: Vision Transformer, published at ICLR 2021 [43].”
MethodSharp-NeRF: Grid-based Fast Deblurring Neural Radiance Fields using Sharpness Prior
Lee, B., Lee, H., Ali, U., & Park, E. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00367 · Cited by 24
“The calculation of depth information is accomplished through the application of the MIDAS [30].”
MethodPanboonyuen, T., Thongbai, S., Wongweeranimit, W., Santitamnont, P., Suphan, K., & Charoenphon, C. (2021) · Information · doi:10.3390/info13010005 · Cited by 23
A Survey on Monocular Re-Localization: From the Perspective of Scene Map Representation
Miao, J., Jiang, K., Wen, T., Wang, Y., Jia, P., Wijaya, B., Zhao, X., Qian, C., Xiao, Z., Huang, J. H., et al. (2024) · IEEE Transactions on Intelligent Vehicles · doi:10.1109/tiv.2024.3378716 · Cited by 27
InvPT++: Inverted Pyramid Multi-Task Transformer for Visual Scene Understanding
Ye, H., & Xu, D. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3397031 · Cited by 30
“Recently, transformer models have been introduced to capture long-range spatial relationships in dense prediction problems [11], [12].”
BackgroundMethodHow NeRFs and 3-D Gaussian Splatting Are Reshaping SLAM: A Survey
Tosi, F., Zhang, Y., Gong, Z., Sandström, E., Mattoccia, S., Oswald, M. R., & Poggi, M. (2026) · IEEE Transactions on Robotics · doi:10.1109/tro.2026.3666139 · Cited by 109
“Then, enhancing these sparse cues involves a dense geometry module predicting dense depth maps using the state-of-the-art monocular depth network, DPT [128], [148], and extracting surface normals using Omnidata [123] from monocular RGB input.”
MethodFishTrack23: An Ensemble Underwater Dataset for Multi-Object Tracking
Dawkins, M., Prior, J. H., Lewis, B., Faillettaz, R., Banez, T., Salvi, M., Rollo, A., Simon, J., Campbell, M. D., Lucero, M., et al. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00701 · Cited by 22
“…integrated, including a separate motion fork that’s fused in later network layers instead of directly at the input [13], unsupervised pre-training over motion channels, 3D convolution in base layers of temporally stacked images, recurrent neural networks, or vision transformer-based methods [29].”
BackgroundPanelNet: Understanding 360 Indoor Environment via Panel Representation
Yu, H., He, L., Jian, B., Feng, W., & Liu, S. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00091 · Cited by 20
“ViT [5] and its following works [22, 23, 27, 39] demonstrate that Transformers are suitable for capturing long-range dependencies for vision tasks by achieving superior results against CNN based models on image classification, image segmentation and dense prediction.”
BackgroundSelfSplat: Pose-Free and 3D Prior-Free Generalizable 3D Gaussian Splatting
Kang, G., Yoo, J., Park, J., Nam, S., Im, H., Shin, S.-H., Kim, S., & Park, E. (2025) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52734.2025.02050 · Cited by 34
“We build our feature fusion block with Dense Prediction Transformer (DPT) [40] module.”
A Rotation-Translation-Decoupled Solution for Robust and Efficient Visual-Inertial Initialization
He, Y.-J., Xu, B., Ouyang, Z.-P., & Li, H. (2023) · Computer Vision and Pattern Recognition · doi:10.1109/cvpr52729.2023.00078 · Cited by 33
“Benefiting from deep learning-based monocular depth estimation [32,33], Zhou et al.”
Multi-Camera Collaborative Depth Prediction via Consistent Structure Estimation
Xu, J., Liu, X., Bai, Y., Jiang, J., Wang, K., Chen, X., & Ji, X. (2022) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3503161.3548394 · Cited by 20
“With the great breakthrough of the transformer network structure in computer vision, [27, 35] adopts the model related to the transformer, and have made great progress in many datasets.”
BackgroundAn edge-guided method to fruit segmentation in complex environments
Sheng, X., Kang, C., Zheng, J., et al. (2023) · Computers and Electronics in Agriculture · doi:10.1016/j.compag.2023.107788 · Cited by 20
Wei, R., Li, B., Zhong, F., Mo, H., Dou, Q., Liu, Y., & Sun, D. B. (2024) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2024.3409392 · Cited by 19
“DPT [30] is a monocular depth estimation method that adopted the vision transformer as a backbone.”
AI Flow: perspectives, scenarios, and approaches
An, H., Hu, W., Huang, S., Huang, S., Li, R., Liang, Y., Shao, J., Song, Y., Wang, Z., Yuan, C., et al. (2026) · Vicinagearth. · doi:10.1007/s44336-025-00031-y · Cited by 25
EGA-Depth: Efficient Guided Attention for Self-Supervised Multi-Camera Depth Estimation
Shi, Y., Cai, H., Ansari, A., & Porikli, F. (2023) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw59228.2023.00017 · Cited by 20
“By leveraging deep learning, researchers have been able to achieve significantly more accurate depth estimation [1, 12, 14, 38, 51, 54].”
BackgroundChiTransformer: Towards Reliable Stereo from Cues
Su, Q. F., & Ji, S. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.00198 · Cited by 21
“Fusion block: Our convolutional decoder follows the refinement block in [45, 51].”
BackgroundMethodInfluentialTrap Attention: Monocular Depth Estimation with Manual Traps
Ning, C., & Gan, H. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00487 · Cited by 25
Self-supervised monocular depth estimation based on combining convolution and multilayer perceptron
Zheng, Q., Yu, T., & Wang, F. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.105587 · Cited by 19
Salient Object Detection in RGB-D Videos
Mou, A., Lu, Y., He, J., Min, D., Fu, K., & Zhao, Q. (2024) · IEEE Transactions on Image Processing · doi:10.1109/tip.2024.3498326 · Cited by 24
“To this end, we consider generating synthetic depth across the entire RDVS using DPT [90].”
MethodSimple and Effective Synthesis of Indoor 3D Scenes
Koh, J. Y., Agrawal, H., Batra, D., Tucker, R. P., Waters, A., Lee, H., Yang, Y., Baldridge, J., & Anderson, P. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i1.25199 · Cited by 31
3D Cinemagraphy from a Single Image
Li, X., Cao, Z., Sun, H., Zhang, J., Xian, K., & Lin, G. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00446 · Cited by 27
“For a fair comparison, all methods utilize the depth maps estimated by DPT [45].”
BackgroundMethodInfluentialPixel shuffling is all you need: spatially aware convmixer for dense prediction tasks
Ibrahem, H., Salem, A., & Kang, H. (2024) · Pattern Recognition · doi:10.1016/j.patcog.2024.111068 · Cited by 19
PolyMaX: General Dense Prediction with Mask Transformer
Yang, X., Yuan, L., Wilber, M. J., Sharma, A., Gu, X., Qiao, S., Debats, S. R., Wang, H., Adam, H., Sirotenko, M., et al. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00109 · Cited by 19
“Entering the deep learning era [37, 46, 82], enormous efforts have been made to tackle dense prediction problems, including but not limited to, image segmentation [15–17, 62, 112], depth estimation [28, 31, 48, 67, 76, 77], surface normal prediction [27,44,65,91], and others [1,23,26,56, 66, 108].”
BackgroundHuang, Z., Jampani, V., Thai, A., Li, Y., Stojanov, S., & Rehg, J. M. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.01241 · Cited by 21
“5D estimation that generalizes to various scenes [7, 34, 35], we propose to use off-the-shelf surface normals as additional geometric supervision for our task.”
MethodQiu, S., Zaheer, Q., Shah, S. M. A. H., Shah, S. F. H., Ehsan, H., Atta, Z., Ai, C., Wang, J., Wang, W., & Peng, J. (2024) · Measurement · doi:10.1016/j.measurement.2024.116557 · Cited by 20
Dense Contrastive-Based Federated Learning for Dense Prediction Tasks on Medical Images
Yang, Y., Liu, X., Gao, T., Xu, X., Zhang, P., & Wang, G. (2024) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2024.3357947 · Cited by 20
“Dense prediction for image is the task of predicting a label at pixel-level [20], which means that region of interests (RoIs) are a dense sample of possible bounding boxes around points of interest (image or feature maps).”
Background3D reconstruction from structured-light profilometry with dual-path hybrid network
Wang, L., Lu, D., Qiu, R., et al. (2022) · EURASIP Journal on Advances in Signal Processing · doi:10.1186/s13634-022-00848-5 · Cited by 19
“A pure transformer or swin transformer network has an obvious advantage in the global perception of the pattern, but the pattern detail information is lost in the division of patches [45].”
BackgroundHierarchical Visual Primitive Experts for Compositional Zero-Shot Learning
Kim, H., Lee, J., Park, S., & Sohn, K. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00522 · Cited by 20
“Following [51], we tile p o to match the spatial dimension with Z np and concatenate them to obtain an object-contextualized feature As shown in Fig.”
MethodUnsupervised Camouflaged Object Segmentation as Domain Adaptation
Zhang, Y., & Wu, C. (2023) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw60793.2023.00468 · Cited by 27
“Though improvement [71, 36] has been made as the booming development of vision transformers [13, 42], this setting requires either dense labels (i.”
BackgroundRobust Reflection Removal With Flash-Only Cues in the Wild
Lei, C., Jiang, X., & Chen, Q. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3314972 · Cited by 19
“We first estimate the depth of the transmission and reflection through a pretrained depth estimation network DPT [51].”
MethodVQ3D: Learning a 3D-Aware Generative Model on ImageNet
Sargent, K., Koh, J. Y., Zhang, H., Chang, H., Herrmann, C., Srinivasan, P. P., Wu, J., & Sun, D. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00391 · Cited by 25
“We employ the pretrained depth prediction trans-former model DPT [30] which produces pseudo-GT inverse depth estimates for the images in our training datasets.”
MethodJoint learning of frequency and spatial domains for dense image prediction
Jia, S., & Yao, W. (2022) · ISPRS Journal of Photogrammetry and Remote Sensing · doi:10.1016/j.isprsjprs.2022.11.001 · Cited by 24
ZeroShape: Regression-Based Zero-Shot Shape Reconstruction
Huang, Z., Stojanov, S., Thai, A., Jampani, V., & Rehg, J. M. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.00959 · Cited by 21
“We used a shared DPT [42] backbone for the depth and camera estimator, and use two different shallow heads to predict D from the local tokens and K from the global token.”
Method
Metadata
- Authors
- Ranftl, R., Bochkovskiy, A., & Koltun, V.
- Affiliations
- Intel Labs
- Intel (United States) (US)
- Venue
- 2021 IEEE/CVF International Conference on Computer Vision (ICCV)
- Publisher
- IEEE
- Published
- 2021-10
- Pages
- 12159-12168
- OpenAlex
- W3136635488
- Cited by
- 2,373
- References
- 88
- Citations by year
- 2026: 480 · 2025: 607 · 2024: 586 · 2023: 431 · 2022: 225 · 2021: 44
- FWCI
- 104.94
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE/CVF International Conference on Computer Vision (ICCV) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- TIB Data Manager
- IEEE/CVF International Conference on Computer Vision (ICCV)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Vision and Imaging
- Keywords
- vision transformer, dense prediction, monocular depth estimation, semantic segmentation, convolutional decoder, global receptive field, Pascal Context, NYUv2, KITTI, ADE20K
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
