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 1,601–1,700 of 2,248
TAJ: A Platform for Integrating Tumor Detection and Depth Perception for Endoscopic Surgery
Fakhry, A., Le, T., Zhang, J., & Li, R. (2024) · International Symposium on Medical Robotics · doi:10.1109/ismr63436.2024.10585947
“The output of the MiDaS algorithms are depth maps, color-coded based upon the proximity of each region from the perspective of the camera [14].”
MethodMLA3DVG: Dual-modality multi-level semantic alignment for robust monocular 3D visual grounding
Chen, S., Xu, Z., Wei, L., Chen, Z., Wen, G., Lang, C., & Song, B. (2026) · Pattern Recognition · doi:10.1016/j.patcog.2026.114622
Wang, G., Wang, Z., & Yang, K. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37369-4_20
RGB-Derived Canopy Height Models for Riparian Woody Vegetation Monitoring Using Depth Anything V2
Choi, H., An, S., Lee, C., Cho, K., & Seong, B. (2026) · Remote Sensing · doi:10.3390/rs18183063
McMurry, S., Goldstein, B. R., Alyetama, M., & Kays, R. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.06.14.732225
Dou, B., Ma, Y., Zhang, T., & Yuan, Z. (2025) · International Conference on Multimedia Retrieval · doi:10.1145/3731715.3733401
“Both few-shot[21] and zero-shot[6, 18, 27, 32, 44] neural segmentation approaches are usually inspired by the development of pre-trained semantic-aware feature extractors (such as DINO-ViT[25]) or segmenters (such as SAM[17] or MaskDINO[20]), as RD-NeRF[21], DFF[18] distill semantic-aware features into scene represent…”
InfluentialLayerLock: Non-Collapsing Representation Learning with Progressive Freezing
Erdogan, G., Parthasarathy, N., Ionescu, C., Hudson, D. A., Lerchner, A., Zisserman, A., Sajjadi, M. S. M., & Carreira, J. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.01809
Linux AGX: An Adaptive GPU eXtension to Linux Fair Scheduling for Physical AI and Robotic Systems
Shirvani, S., & Liu, C. (2026) · ACM SIGOPS Symposium on Operating Systems Principles (SOSP) · doi:10.1145/3830418.3843893
3D Visualization from Single-Chip Camera in Single Port Access Fetal Robotic Surgery
Li, Y., Li, X., Zhang, X., et al. (2025) · International Symposium on Medical Robotics · doi:10.1109/ismr67322.2025.11025961
“While metric monocular depth estimation has been investigated in medical applications such as laparoscopy [17] using trained DPT models [18] on laparoscopic datasets, spina bifida surgery presents unique challenges.”
Depthtective: A depth-aware framework for spatio-temporal deepfake detection
Bisogni, C., Costante, L., Loia, V., Nappi, M., & Pero, C. (2026) · Array · doi:10.1016/j.array.2026.101278
Enhancing the Safety Horizon: Robust Uncertainty Quantification for Monocular Autonomous Navigation
Estrera, D. J., & Azcarraga, J. J. (2026) · IEEE Annual Computers, Software, and Applications Conference (COMPSAC) · doi:10.1109/compsac69091.2026.00489
“C. Geometric Uncertainty (MiDaS + TTA) MiDaS_small [5] is used.”
Improving 3D Reconstruction from Limited Samples: A Depth-Guided NeRF Approach
Liao, L., Chen, J., Zhang, Z., Huai, X., Zhang, C., Ren, X., & Zhang, B. (2024) · IEEE Information Technology, Networking, Electronic and Automation Control Conference (ITNEC) · doi:10.1109/itnec60942.2024.10733017
“This model is an advanced depth estimation system that combines a DINOv2 [8] encoder with a DPT [9] decoder.”
MethodEndoDA3: Unified Depth and Pose Estimation for Endoscopic Video
Tan, J., Li, Z., Tuo, H., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38236-8_15
Stereo-Knowledge Distillation from dpMV to Dual Pixels for Light Field Video Reconstruction
Garg, A., Mallampali, R., Joshi, A., Govindarajan, S., & Mitra, K. (2024) · IEEE International Conference on Computational Photography (ICCP) · doi:10.1109/iccp61108.2024.10644854
“4 that methods incorporating dark knowledge out-perform purely dual-pixel solutions [3] as well as massively trained monocular solutions [15], [16], [17], particularly in challenging foreground-background separation where guidance from dual pixels proves indispensable.”
BackgroundMethodInfluentialAddressing Low-Shot MVS by Detecting and Completing Planar Surfaces
Kataria, R., Li, Z., DeGol, J., et al. (2024) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv62453.2024.00015
“Recently, vision transformers [12] have been used in semantic segmentation networks [25, 26, 34, 43] where they have demonstrated the ability to capture long-range dependencies.”
BackgroundSmart Virtual Fence Surveillance for Precision Intrusion Detection
Lin, P.-J., Lam, K.-H., Wu, C.-S., et al. (2026) · International Conference on Control, Automation and Robotics (ICCAR) · doi:10.1109/iccar69571.2026.11549769
“Subsequent studies, such as the Dense Prediction Transformer (DPT) [16] and Learning from Relative Depth and Semantics (LeReS) [17], have further elevated the structural quality of depth estimation through global attention mechanisms and semantic guidance.”
NoDrift3R: Raymap-Guided Coupling for Drift-Robust Unposed Feed-Forward 3D Reconstruction
Sun, X., Liu, L., Yang, S., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37435-6_24
Recent Researches on Media Engineering
Arai, H., Tarashima, S., Kawamura, K., Hori, A., Tada, M., Mochizuki, T., & Ogawa, T. (2022) · The Journal of The Institute of Image Information and Television Engineers · doi:10.3169/itej.76.78
RGB-Assisted Depth Completion for LiDAR Occlusion in Manufacturing Environments
Hwang, J. H., Ha, S. Y., Choi, J. D., et al. (2026) · International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA) · doi:10.1109/acdsa67686.2026.11467895
“Methods such as MiDaS, DPT, and AdaBins [12], [13] leverage deep learning to infer dense depth maps from single images.”
Semisupervised Video-SAR Multishadow Tracking via Spatiotemporal Dynamics
Yang, X., Gao, S., Zhang, B., et al. (2026) · IEEE Transactions on Aerospace and Electronic Systems · doi:10.1109/taes.2026.3710453
“…frameworks, such as vision transformers and DINOv2 [18], [19], demonstrate powerful feature representation capabilities, and decoders incorporating atrous spatial pyramid pooling (ASPP) [20], [21] excel in multiscale perception, they rely heavily on massive, fine-grained pixel-level annotations.”
A Two-Branch Framework for Monocular Depth Estimation
Zhou, D., Zhang, M., & Li, B. (2023) · International Conference on Image, Vision and Computing (ICIVC) · doi:10.1109/icivc58118.2023.10270662
“(a) Color image, (b) Dpt [29], (c) Bts [14], (d) Adabins [13], (e) Ours”
BackgroundMethodInfluentialHan, J., Li, M., Zhang, Z., Tan, W., & Wang, Y. (2026) · Computerized Medical Imaging and Graphics · doi:10.1016/j.compmedimag.2026.102824
Zhang, J., Luo, Y., & Yang, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-5737-0_35
Depth-aware Trimodal Network for RGB-T Salient Object Dection
Zhang, Z., & Li, W. (2022) · International Conference on Electronic Information Technology and Computer Engineering (EITCE) · doi:10.1145/3573428.3573476
“The quality of the depth map synthesized by DPT [3] is significantly higher than that of Megadepth [14].”
ResultTowards Reliable Advertising Image Generation Using Human Feedback
Du, Z., Feng, W., Wang, H., Li, Y., Wang, J., Li, J., Zhang, Z., Lv, J., Zhu, X., Jin, J., et al. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72661-3_23 · Cited by 6
Intra-Class Patch Swap for Self-Distillation
Choi, H., Jeon, E. S., Shukla, A., & Turaga, P. K. (2025) · Neurocomputing · doi:10.1016/j.neucom.2025.130408
Toderici, D., Rodanciuc, T.-G., Micu, G.-A., et al. (2026) · Electronics · doi:10.3390/electronics15153295
Foundation-Prior-Guided Target Scene Adaptation for Underwater Stereo Metric Depth EstimationPreprint
Huang, K., Liu, Y., Liu, J., et al. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7227215
Plug-and-play adapter for fusing convolutional neural network with vision transformer
Chen, B., Fan, X., & Wu, S. (2024) · Journal of Electronic Imaging · doi:10.1117/1.jei.33.5.053050 · Cited by 1
Probing the 3D Object-Level Understanding of Pre-trained Detection Transformers
Kim, R., Samplawski, C., & Marlin, B. M. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37035-8_6
Region Guide Point for Depth Estimation
Jun, W., & Tseng, K. K. (2023) · Smart innovation, systems and technologies · doi:10.1007/978-981-99-0105-0_3
Park, B. (2024) · Journal of Sensor Science and Technology · doi:10.46670/jsst.2024.33.2.98
Leveraging Mono-Modal Data to Generate Multimodal Representations for Industrial Anomaly Detection
Achkar, C. E., Makhoul, A., & Azar, J. (2026) · IEEE Annual International Computer Software and Applications Conference (COMPSAC) · doi:10.1109/compsac69091.2026.00024
“Depth sensing and monocular depth estimation add geometric information useful for identifying misalignments, dents, and surface deformations, with MiDas commonly used for depth prediction from RGB images [3], [22], [23].”
Label-Efficient Semi-Supervised Building Semantic Segmentation for High-Resolution UAV Imagery
Lee, G., Baek, J., & Han, Y. (2026) · Remote Sensing · doi:10.3390/rs18183250
Joint depth estimation and multi-model SLAM for robust perception in structure-degraded environments
Wang, W., Ji, W., Xiao, J., Hu, X., Jia, Z., & Shi, J. (2026) · The Journal of Supercomputing · doi:10.1007/s11227-026-08461-1 · Cited by 1
Lu, H., An, Y., Sun, B. C., et al. (2026) · International Conference on Computer and Communication Systems (ICCCS) · doi:10.1109/icccs69761.2026.11613519
“DynaSLAM [7] restores occluded areas via multi-view fusion, while learning-based depth completion [21]–[23] improves completeness at high computational cost.”
Zhang, Y., ZHANG, J., Xu, W., & Hashimoto, K. (2026) · Biomimetic Intelligence and Robotics · doi:10.1016/j.birob.2026.100371
Diversity-Aware View Partitioning for Scalable VGGT
Park, J., Choi, D., Seo, A., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37023-5_30
A 3D Reconstruction Network Based on Multi-sensor
Zhou, Y., Lv, J., Ma, Y., & Ma, X. (2022) · Communications in computer and information science · doi:10.1007/978-981-19-9198-1_44
Ryu, J. H., Cha, S. W., & Jeon, E. S. (2026) · Biomimetics · doi:10.3390/biomimetics11090665
“However, most existing studies assume a wide and unobstructed field of view (FoV), which is an idealized setting where the full context of a driving scene is available [3,4].”
Learning to Generate and Predict New Conditional Small Organic Molecules
Alsowat, A., Allahyani, R., Alsulami, R., Almaqati, S., Alafif, T., Hawsawi, M. B., Alsaidi, H., & Alsubait, T. (2024) · IEEE International Conference on Computing, Engineering and Design (ICCED) · doi:10.1109/icced64257.2024.10982921 · Cited by 1
“Our work is inspired by the success of WGAN and ViT in image generation [10], [14] and image prediction [11], [15], respectively.”
Deep learning-based segmentation of human oocytes with cross-dataset evaluation
Lei, Z., Xu, X., Tilia, L., du Toit-Thompson, S., Horta, F., Gilchrist, R. B., Walls, M. L., Bhargava, A., Ledger, W., & Goldys, E. M. (2026) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2026.109507
Converting 2D Images to Point Cloud Using Depth Estimation
Thukral, S., Shetty, T., Das, A., Tiwari, V., & Kaur, R. (2025) · Lecture notes in networks and systems · doi:10.1007/978-981-96-5607-3_34
GazeVAE: Gaze Visual Attention Estimator
Lin, B.-L., Lin, Y.-X., Mamade, B. K., et al. (2024) · International Conference on Consumer Electronics · doi:10.1109/icce59016.2024.10444171
Drive and Perception Systems of Flexible Manipulators
Qu, J. (2026) · Springer eBooks · doi:10.1007/978-981-95-6232-9_3
MSTG: Multi-Scale Transformer with Gradient for joint spatio-temporal enhancement
Lin, X., Chen, J., Ai, S., Liu, J., Li, B., Li, Q., & Ma, R. (2024) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2024.104209
Event-Based Desnowing for Autonomous Driving
Muglikar, M., Messikommer, N., Cannici, M., et al. (2026) · IEEE Transactions on Robotics · doi:10.1109/tro.2026.3697155
“1: Compute depth map D ( x ) from J ( x ) using a depth estimation model [47] 2: Generate haze image I haze ( x ) using D ( x ) „ haze parameters A , and atmospheric scattering model [46] 3: Overlay snow events on I haze ( x ) to obtain snow-occluded image Z ( x ) 4: Initialize synthetic event…”
INSPIRED: INdoor Scene Placement with Inpainting, Style Recognition and Estimation of Depth
Pai, S. R., Narayan, S., Sadalgekar, S. U., Shivanna, B., & Shetty, U. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-31144-3_15
Globally Aware and Locally Sensitive 360° Depth Estimation
Zhang, Q. (2023) · International Conference on Information Technology in Medicine and Education · doi:10.1109/itme60234.2023.00109
“In contrast, Vision Transformers [10] outperform CNNs in remote feature learning. however, they have difficulties in learning local structure, especially for panoramic images with severe spatial variation.”
BackgroundAdjacent-view Transformers for Supervised Surround-view Depth Estimation
Guo, X., Yuan, W., Zhang, Y., Tian, Y., Zhang, C., Zhu, Z. D., Zou, Q., & Chen, L. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11246758 · Cited by 5
“Recently, the attention mechanism has already proven to be effective in various computer vision tasks, such as image classification [41], object detection [68] and depth estimation [43], [44], [45], [3], [69].”
Dual-Scene Representation Learning for Traffic Dehazing and Reconstruction
Zhang, B. H., Ma, J., & Liu, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-5737-0_17
A Multi-Modal Benchmark for Long-Range Depth Evaluation in Adverse Weather Conditions
Walz, S., Ramazzina, A., Scheuble, D., Brucker, S., Zuber, A., Ritter, W., Bijelic, M., & Heide, F. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11246099
“In particular, we use DPT [39] as monocular RGB depth estimation model, CREStereo [40] as stereo approach, CompletionFormer [13] as LiDAR depth completion method, Gated Stereo [20] for (stereo) gated depth estimation and Gated-RCCB-Stereo [21] for cross-spectral stereo depth estimation.”
Udugama, U. V. B. L., George Vosselman, A., & Nex, F. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11246974
“In the following sections, we delve into the design details of the WMCA and GGFM blocks.”
Magdy, D., AbdelRaouf, A., & Adel, S. (2025) · Intelligent Methods, Systems, and Applications (IMSA) · doi:10.1109/imsa65733.2025.11166813
“Recent ViT-based depth estimators like DPT [12] offer promising results in dense prediction with global coherence.”
Deep learning-based computer vision in forest monitoring and management: a systematic review
Forkuo, G. O., & Borz, S. A. (2026) · Biodiversity and Conservation · doi:10.1007/s10531-026-03412-x
“Dense-prediction and deep image transformers (Ranftl et al. 2021; Touvron et al. 2021 ; Peng et al. 2021), transformer-based detection frameworks (Misra et al. 2021; Vaidwan et al. 2021 ; Ma et al. 2021 ), and classical anchor-based detectors (Redmon and Farhadi 2018 ) have been comprehensively…”
3D Reprojection-Driven Robot Navigation Improves Depth Sensing
Chancán, M., Wong, A., & Abraham, I. (2025) · International Conference on Advanced Robotics and Mechatronics (ICARM) · doi:10.1109/icarm65671.2025.11293600
Vision-Based Cooperative Train–Ground Localization with Monocular Metric Distance Regression
Wang, Y., He, Z., Xie, J., Zhang, J., Xu, N., Zhao, J., & Fan, K. (2026) · International Conference on Artificial Intelligence and Electromechanical Automation (AIEA) · doi:10.1109/aiea70743.2026.11632981
“Early methods relied on geometric cues and handcrafted features, while recent advances have been driven by deep learning models based on convolutional neural networks [1] and transformer architectures [2].”
MS3D: High-Quality 3D Generation via Multi-Scale Representation Modeling
Luo, G., & Zhang, J. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.02444 · Cited by 1
“Inspired by successful multi-scale methods in 2D vision tasks [50, 51], we develop a multi-scale image feature extraction and scale-corresponding feature integration mechanism that works in perfect synergy with our hierarchical structured latent representation.”
HKAFusion: Hybrid Kernel Attention Network for Medical Image Fusion
Zeng, S., Liu, H., Yao, C. L., Sun, J., Tao, S., & Ban, X. (2025) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm66473.2025.11357018
“To achieve a unified representation of the advantages offered by multiple modalities, medical image fusion [6] has emerged as a crucial technique.”
PSGS: Text-Driven Panorama Sliding Scene Generation via Gaussian Splatting
Zhang, X. Y., Chen, S., Jiale, Z., et al. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11461817
Yiquan, 吴. W., & Haobo, 谢. X. (2025) · Acta Optica Sinica (Online) · doi:10.3788/aosol250454
Gong, Z., Guo, X., Hu, W., et al. (2026) · IEEE Transactions on Cloud Computing · doi:10.1109/tcc.2026.3670345
VFAvatar: Feed-Forward 3D Avatar Reconstruction from Casual Image Collections
Huang, S., Wang, Z.-X., Qin, X., et al. (2026) · International Conference on Computer Graphics and Interactive Techniques · doi:10.1145/3799902.3811171
An, Z., Peng, L., Chen, S., et al. (2026) · Stochastic Environmental Research and Risk Assessment · doi:10.1007/s00477-026-03185-3
Enhancing neural radiance fields with geometry-aware transformers and depth fusion
Xu, M., Bai, Z., Cao, C., et al. (2026) · Multimedia Systems · doi:10.1007/s00530-025-02193-7
“While monocular depth estimation has achieved strong accuracy and robustness through multi-scale feature aggregation and geometric priors [ 14 , 15], its integration into NeRF-based rendering pipelines has been relatively limited [16, 17 ].”
Beram, S. R., Lalchhanhima, R., & Singh, K. R. (2026) · Evolving Systems · doi:10.1007/s12530-026-09825-x
“Vision transformers capture the global context effi - ciently for decision making but they need large datasets to avoid overfitting issue Rene et al. ( 2021).”
STF-Diff: The key to image style transfer using extended style inversion is the style token filter
Zhao, H., Zhang, J., Yang, Q., Wang, C., Pu, Y., & Zhao, Z. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.116777
A Hybrid Shadow-Aware Network for Transparent Object Detection and Transparency Estimation
Vaidya, C. D., Budhe, S., Paighan, U., et al. (2025) · doi:10.1109/icft66708.2025.11336341
“[12] used U-Net with pyramid attention for TOD, resulting in better feature aggregation.”
MCID: Multi-aspect Copyright Infringement Detection for Generated Images
Huang, C., Jia, Z., Fei, H., Zhu, Y., Yuan, Z., Deng, Y., Zhang, J., Duan, X., Zhang, J., & Zhou, J. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.01499
“We use DPT [37] to extract the depth map of x as structural information guidance.”
Gorshenin, A. K., & Dostovalova, A. M. (2025) · Doklady Mathematics · doi:10.1134/s1064562425700255
“…the MMRFiGN model outperformed the results of semantic segmentation of small (cars) and large objects in terms of the -score of both transformer (Dense Prediction Transformer (DPT) [17], Seg-Former [18], LWGANet [19]) models and convolutional (U-Net [20], ENet [21], DeepLabV3 [22]) architectures.”
Unsupervised brain MRI image registration based on 3D hybrid ViT and convolutional U-netPreprint
Wang, J., Liu, F., Shu, S., Tao, G., & Zhou, F. (2023) · Research Square · doi:10.21203/rs.3.rs-2898722/v1
FSDE-net: Coarse-to-Fine Feature Separation Depth Estimation Network
Ge, X., & Guo, F. (2025) · doi:10.1109/iscait64916.2025.11010465
“At the same time, new architectures based on vision Transformer (ViT) [11], such as DPT [12], show better performance than traditional CNNs in complex scenarios through global attention modeling.”
Recurrent alignment with dynamic query attention transformer framework for rainy video restoration
Shaikh, J. A. H., Mukane, S., & Randive, S. (2026) · Imaging Science Journal · doi:10.1080/13682199.2026.2624235
“To show the practical benefit, we ran a modern depth estimator [63] and the Segment Anything Model (SAM) [64] on both degraded frames Qualitative results analysis of the proposed method with SOTA RMFD [8], DLF [46], ViWS-Net [53], DFST [51] methods on deraining with veiling effect on real-world…”
NVS-HO: A Benchmark for Novel View Synthesis of Handheld Objects
Ali, M., Carranza-García, M., Fioraio, N., Salti, S., & Stefano, L. D. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31583-0_37
Enhancing Monocular Depth Estimation Using Attention in Hybrid Network
Huo, Z., & Zhao, L. (2023) · International Conference on Wireless Communications and Signal Processing (WCSP) · doi:10.1109/wcsp58612.2023.10404542
“In contrast, methods based on ViTs model long-range dependencies and global information, enabling better capturing of contextual relationships in images [5], [7].”
BackgroundMethodGupta, M., Devi, B., & Shankar, V. G. (2025) · International Conference on Intelligent Computing, Information and Control Systems (ICOIICS) · doi:10.1109/icoiics67115.2025.11390199
“Transformer-based models, notably DPT [3], leveraged self-attention to capture global context, improving performance at the expense of computational complexity.”
Enhancing Monocular Depth Estimation Using an Enhanced Encoder--Decoder ModelPreprint
El-Alfy, Y. E., Baroudi, U. A., & Luqman, H. (2023) · doi:10.36227/techrxiv.23995554.v1
Zhang, M., & Zheng, Y. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11461794
Vision on the Warehouse Floor: Perceptive HRI for Autonomous Forklifts
Jung, M., Hempel, T., Schulz, P., & Al-Hamadi, A. K. (2026) · Robotics · doi:10.3390/robotics15100191
Underwater Salient Object Detection Based on Swintransformer
Jiao, Z., Wang, R., Fu, B., Zhang, T., & Zhang, N. (2025) · Communications in computer and information science · doi:10.1007/978-981-96-1614-5_15
Improved APF-Based Obstacle Avoidance with Monocular Depth Estimation
周炳恒, Su, J., & Yang, J. (2026) · Chinese Control and Decision Conference (CCDC) · doi:10.1109/ccdc69976.2026.11560556
“( sucas@seu.edu.cn ) State-of-the-art transformer-based models such as DPT [3], and Depth Anything [4] have demonstrated exceptional cross-dataset generalization capabilities through large-scale pre-training.”
A Multi-Scale Global Fusion-Based Method for Surface Fissure Extraction from UAV Imagery
Zhou, M., Ji, M., Jin, F., et al. (2026) · Sensors · doi:10.3390/s26051440
“Research indicates that CNNs provide more stable representations for anisotropic and narrow structures [29].”
Ohshima, Y. (2025) · Taiikugaku kenkyu (Japan Journal of Physical Education Health and Sport Sciences) · doi:10.5432/jjpehss.015-24089
Chen, L., Chen, X., & Cai, S. (2026) · doi:10.1109/aetcse69203.2026.11504313
“The original visual token output by the encoder is then reassembled and converted into multi-scale feature maps of different resolutions [18].”
Li, F., Lu, X., Shen, J., et al. (2026) · Land · doi:10.3390/land15081402
Furnit.Ar: AI-Powered Augmented Reality Interior Designing Application
Dhar, S. U., Haldar, S., Pete, S., et al. (2025) · IEEE International Conference on Computer Vision and Machine Intelligence (CVMI) · doi:10.1109/cvmi66673.2025.11337706
“Their proprietary backend likely employs a combination of techniques: – Depth Estimation: Predicting depth maps from images using models like MiDaS [33] or DPT [34].”
Monocular 3D Reconstruction Based on Deep Convolutional Neural Networks
Chen, D., & Dai, Z. (2025) · International Conference on Image, Vision and Computing (ICIVC) · doi:10.1109/icivc66358.2025.11200281
“More recent approaches, such as the Vision Transformer – based DPT (Dense Prediction Transformer), utilize global attention to model long-range dependencies across scales, further improving fine structure prediction [10].”
InfluentialScaling Dense Prediction with Latent Decoding
Wu, X., Lao, Y., Wang, C., Yang, S., Zhang, Y., & Zhao, H. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37225-3_3
FuseDPT: Multi-scale and Multi-projection Model for Learning Depth in 360$$^\circ $$C
Paula, M. D. O. P., İmamoğlu, N., Caron, G., & André, A. N. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31666-0_23
Multi-View Depth Generation Method Using ViT-Based Depth Estimation and MVS-Based Depth Optimization
Park, J.-H., Jeong, J.-B., Choi, J.-Y., et al. (2024) · Journal of Broadcast Engineering · doi:10.5909/jbe.2024.29.6.931
Cryo-EM as latent structural landscape microscopyPreprint
Dai, H., Chen, Q., Li, L.-Q., et al. (2026) · bioRxiv · doi:10.64898/2026.04.10.717737
“We leverage odd-even image pairs based on the Noise-to-Noise principle [16] and integrate CryoUNI with the Dense Prediction Transformer (DPT) [29] framework to construct a multi-scale denoising network.”
Machkour, Z., Ortíz-Arroyo, D., & Durdevic, P. (2024) · International Conference on Robotics and Automation Engineering (ICRAE) · doi:10.1109/icrae64368.2024.10851605
“• DPT [17]: A transformer-based model excelling across depth benchmarks.”
Argus: Metric Panoramic 3D Reconstruction for Indoor Scenes
Li, X., Li, L., Wu, Y., Rao, T., Zhang, K., Hui, X., & Pan, C. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37369-4_34
Distance Estimation of Body Joints Position for Human Activities Using Monocular Camera
Oktavian, M., Suratman, F. Y., & Istiqomah, I. (2025) · doi:10.1109/softt67007.2025.11212962
“MiDaS uses an encoder-decoder transformer model in reconstructing an image [6].”
Robust iris segmentation based on ConvNeXts
Hao, L., Zhu, X., Zhang, Q., Chen, K., Wang, J., & Liu, Y. (2023) · Journal of Electronic Imaging · doi:10.1117/1.jei.32.4.043020
A Method for Generating Point Clouds From Monocular Depth Estimation Images
Chavarría-Zamora, L. A., & Soto-Quirós, P. (2024) · IEEE International Conference on BioInspired Processing (BIP) · doi:10.1109/bip63158.2024.10885378
“These tokens are then processed through a series of self-attention and feed-forward layers, allowing the model to learn robust feature representations [7].”
InfluentialA geography of indoors for analyzing global ways of living using computer vision
Mazzarello, M., Klimenka, M., Sanatani, R. P., et al. (2025) · Scientific Reports · doi:10.1038/s41598-025-12198-x
Cai, H., Liu, J., Zhang, J., Xu, J., Zhang, Y., & Yang, Q. (2026) · Computers · doi:10.3390/computers15010019 · Cited by 1
Leveraging Diffusion-Based Augmentation for Robust Semantic Segmentation
Salman, M. M., Gerhardt, C., & Broll, W. (2025) · International Conference on Machine Learning and Applications · doi:10.1109/icmla66185.2025.00154
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
