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 2,101–2,200 of 2,248
Deep Learning-Based 3D Reconstruction from Underwater Gated Images
Zhou, R., Yan, W., & Chen, W. (2026) · International Conference on Control and Automation · doi:10.1109/icca69928.2026.11618036
TNT-Depth: A Transformer-in-Transformer Architecture for Robust Depth Completion
Zhang, Z., & Liang, J. (2025) · China Automation Congress (CAC) · doi:10.1109/cac67268.2025.11487656
Face Anything: 4D Face Reconstruction from Any Image Sequence
Kocasarı, U., Giebenhain, S., Shaw, R., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37255-0_15
Zhu, C., Wan, Y., Wei, D., Wan, X., Xia, P., Huang, X., Wu, Q., & Zhang, Y. (2026) · ISPRS Journal of Photogrammetry and Remote Sensing · doi:10.1016/j.isprsjprs.2026.07.019
Vision Transformers for Thermal Weapon Segmentation
Kambhatla, A., & Ahmed, K. R. (2026) · IEEE SoutheastCon · doi:10.1109/southeastcon63549.2026.11476496
MFOF: Polyp segmentation model based on ordered fusion of multi-scale features
Hu, K., & Li, W. J. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.109165
Realistic Human Image Animation With Dynamic Camera Effects
Kang, M., Kim, J., Yang, C., Kong, K., & Kang, S. (2026) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2026.3692721
“Note that for animated frames V Animation , we utilize a pre-existing monocular depth estimator, DPT [34], to estimate depth for each frame but the following module specifically focuses on estimating the depth of the inpainted background I inpaint .”
InfluentialLearning Compact 3D Gaussians via Feed-Forward Point Fusion
Smart, B., Zheng, C., Laina, I., et al. (2026) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv69130.2026.00059
“We achieve this by adding a new DPT head [46], that we refer to as the ‘Gaussian Feature Head’, to a pretrained VGGT model [59].”
Zhao, M., Wang, S., Xiao, F., Zhang, J., Cheng, X., & Zhu, Y. (2025) · International Conference on Computer Supported Cooperative Work in Design (CSCWD) · doi:10.1109/cscwd64889.2025.11033287
Zhu, Z. G., Zhang, L., Chen, Y., Zhang, T., Wang, Y., & Zhan, M. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-36846-1_7
“Later works improve global reasoning by adopting Vision Transformers [34, 56] or diffusion priors [17, 22, 35], but at the cost of substantially increased model complexity and inference latency.”
3D hand-object reconstruction from RGB images
Swamy, A. K. (2024) · theses.fr (ABES) · doi:10.70675/ee9f3874z733fz4919z87d4z2d174a678024
Enhancing Group Attention for Off-Road Semantic Segmentation via Transition-Aware Refinement
Choi, S.-K., & An, J. (2026) · IEEE Signal Processing Letters · doi:10.1109/lsp.2026.3702520
MRI-based Brain Tumor Segmentation Using Ensemble of Hybrid Deep Learning Models
Kakumani, A. K., Kumar, G. D., Pravallika, A., et al. (2026) · International Conference on Inventive Computation Technologies (ICICT) · doi:10.1109/icict68280.2026.11511034
FEEL (Force-Enhanced Egocentric Learning): A Dataset for Physical Action Understanding
Dessalene, E., He, B., Maynord, M., Tussa, Y., Mantripragada, P., Karabati, Y., Roy, N., & Aloimonos, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37602-2_31 · Cited by 7
AI-Powered Complaint Validation for Vehicle Obstruction in Smart Cities
Raghav, Gupta, D., & Kumar, M. (2025) · International Conference on Computational Intelligence and Communication Networks (CICN) · doi:10.1109/cicn67655.2025.11367831
“The pipeline integrates YOLOv8 for vehicle detection and segmentation detection [3], ResNet for make-model-year (MMY) classification [4], and DPT-Large for depth estimation [1].”
InfluentialBoosting MonoDepth With Foundation Models and Edge Laplace Cross-Entropy in Remote Sensing
Zhou, H., Xiong, W., Liu, W., Li, X., Xiong, B., Zhu, Z., & Rao, Z. (2026) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2026.3664691
“Since deeper ViT tokens encompass richer global contextual information while shallower tokens focus on fine-grained details [34], our method assembles ViT tokens by progressing from the deeper layers to the shallower ones, with progressively increasing resolutions.”
SARG-GS: semantic-augmented and residual-guided 3D Gaussian Splatting for sparse view synthesis
Zhou, H. J., Zhu, H., Qin, J., et al. (2026) · The Visual Computer · doi:10.1007/s00371-026-04646-0
“Depth-aware densi-fication [13, 40, 43], monocular priors from MiDaS [24] and ZoeDepth [2], as well as correspondence estimation using RAFT [29] and LoFTR [28], further enhance structural coherence under sparse views.”
Cho, J., Jang, C., Lee, C., & Kim, K. (2026) · The Journal of Korea Robotics Society · doi:10.7746/jkros.2026.21.2.182
SurfelFlow: Surface-Aware 2D Gaussian Streaming for Monocular Dynamic 4D Reconstruction
Xue, Q., Zhong, Y., Xu, M., et al. (2026) · Applied Sciences · doi:10.3390/app16189288
Bandwidth-Constrained Aerial Edge Perception for Ground Traversability Mapping over Digital Links
Liu, Z., Li, Y., Yong Qiang, J. I. A., Lin, F., Wang, Z., & Yuan, S. (2026) · Electronics · doi:10.3390/electronics15173893
$$\mathtt {M^3VIR}$$ for benchmarking sparse-view novel view synthesis and 3D object removal
Li, Y., Zhou, L., Ling, N., Chen, Z., Wang, W., & Jiang, W. (2026) · Multimedia Tools and Applications · doi:10.1007/s11042-026-21779-5
Liu, P., Sun, H., Liu, Z., et al. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11463411
FuseDepth: Zero-shot metric depth with semantic–geometric fusion of foundation visual priors
Javidnia, H. (2026) · Machine Learning with Applications · doi:10.1016/j.mlwa.2026.101010
Monocular Depth Estimation with Enhanced Target Awareness for UAV Environment Perception
Zhang, L., Sun, J. Y., Zhang, Y., & Li, Z. (2025) · Chinese Control Conference (CCC) · doi:10.23919/ccc64809.2025.11179450
“In recent years, Vision Transformer (ViT)-based meth-ods [21, 22] have been applied to depth estimation tasks.”
Vigdepth: self-supervised monocular depth estimation based on vision GNN and local graph reasoning
Chen, X., & Wang, M. (2026) · International Journal of Machine Learning and Cybernetics · doi:10.1007/s13042-026-03185-3
“Subsequent research has focused on exploring more efficient loss functions [1, 18] or designing more sophisticated network architectures [19, 20].”
Lightweight deformable attention for event-based monocular depth estimation
Yang, J., Wang, S., Wang, J., et al. (2025) · Displays · doi:10.1016/j.displa.2025.103303
Zhou, N., Zhao, X., Zhou, F., Li, J., & Yue, X. (2026) · PLoS ONE · doi:10.1371/journal.pone.0352325
“Moreover, Transformer architectures, originally ubiquitous in natural language processing, are penetrating the segmentation domain, opening novel pathways for superior image comprehension and robust analysis [25–27].”
SACFormer: Unify Depth Estimation and Completion with Prompt
Tang, S., Wu, D., Wang, Y., & Wang, L. (2023) · Lecture notes in computer science · doi:10.1007/978-981-99-8432-9_9
Yeom, S., Son, H., Kang, C., et al. (2025) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2025.3649390
“Preprocessing Components: We use DPT [50] to predict monoculardepthmaps,capturingspatialstructureacrossframes.”
3D Scene Perception for Autonomous Driving
Li, S., Zhou, H., Gao, Y., Cai, X., Yuan, H., & Zhang, W. (2023) · Springer eBooks · doi:10.1007/978-981-99-5439-1_7
Multi-stream information complementarity network for RGB-D camouflaged object detection
Ying, C., Zhou, Z., Li, K., et al. (2025) · The Journal of Supercomputing · doi:10.1007/s11227-025-07201-1
“We benchmarked MICNet against state-of-the-art RGB-D COD models using identical depth estimation models [43, 65].”
Pixel-Depth Prototypical Knowledge Consolidation Network for Deepfake Detection
Asiri, A., Wang, W., Ding, X., Chen, L., & Yu, S. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3015-0_16
Song, L., Jiang, B., & Song, H. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2025.111572
LR 2 Depth: Large-Region Aggregation at Low Resolution for Efficient Monocular Depth Estimation
Ning, C., Xuan, W., Gan, W., & Yokoya, N. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11246436
Image-based Goat Body Size and Weight Estimation via Deep Learning Segmentation and Depth Estimation
Shrestha, S. K., R, H. Y., Theja Sree, M., et al. (2026) · doi:10.1109/icc-cns70518.2026.11606041
“MiDaS (Mixed Data Sampling) treats depth estimation as a dense regression problem [11].”
Improvement of RGB-Based Person Reidentification Using Artificial Modalities
Rayani, M. S., Keshavarz, A., Ghimatgar, H., et al. (2026) · doi:10.1109/ibaaids66771.2026.11567402
TEPR: Thermal Edge-Prompted Geometry Learning for Multi-View 3D Reconstruction
Li, Z., Liu, Z., Wang, W., et al. (2026) · International Journal of Computer Vision · doi:10.1007/s11263-026-02997-8
VRMusicStage: A System for Converting Fixed-Camera Music Stage Videos into Immersive VR Content
Leem, S., Jeong, S., Cho, Y., Lee, Y., & Lee, J. (2025) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3746027.3754563
“ControlNet [47] extends this by incorporating image-based conditions, such as Canny edges [5] or depth maps [2, 30], using dedicated networks.”
DA-YOLO: Depth-Aware YOLO framework for Autonomous Driving Using MiDaS
Santhosh, M., Sibi, P., & C, D. J. (2026) · doi:10.1109/icitsif69060.2026.11609063
“The YOLO backbone first extracts semantic features from the RGB input, while MiDaS Small independently generates multi-scale depth maps from the same image [9], [10].”
Ghaffar, F., Khan, B., Jalali, S. M. J., & Lim, C. P. (2025) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc58881.2025.11343042
“For depth estimation, early works like [18] used coarse-to-fine CNNs, while modern approaches like MiDaS [19] and DPT [20] leverage transformer backbones.”
Self-supervised Monocular Depth Estimation for Dynamic Objects with Ground Propagation
Li, H., Poggi, M., Tosi, F., & Mattoccia, S. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11246123
“Transformer-based models [29], [30] have further improved accuracy, while lightweight architectures [31]–[33] address real-time applications.”
Transformer-Based End-to-End Model for Monocular Visual Odometry on Urban Traffic Scenes
Pontarolo, G., & Todt, E. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-39898-7_33
SD-Human: Stable Dynamic Human Scene Rendering from a Fixed Single-View
Hong, J., Dong-hu, L., Chen, S., Yang, Z., & Li, N. (2025) · International Conference on Computer, Artificial Intelligence and Control Engineering (CAICE) · doi:10.1145/3727648.3727749
“Our proposed outperforms other methods in 𝛿 , AbsRel, Rmse, and R.Log [ 14 ] .”
2K Retrofit: Entropy-Guided Efficient Sparse Refinement for High-Resolution 3D Geometry Prediction
Zhang, T., Liang, Z., Song, Z., Wang, N., Zhang, X., Cai, X., Zhu, Z., Wu, K., Wang, G., & Fan, Z. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37550-6_9
DPCT: Efficient High-Resolution Depth Prediction via Cross-Covariance Attention Transformers
Gan, H., Jin, Y., Li, J., & Wei, K. (2025) · International Conference on Computer Engineering and Application (ICCEA) · doi:10.1109/iccea65460.2025.11103050
“Within this advancement, the Depth prediction Transformer (DPT) [1] emerges as a groundbreaking framework that integrates a ViT-based encoder with a multi-scale feature fusion decoder, achieving state-of-the-art performance in pixel-level prediction tasks.”
Lunar Terrain Depth Mapping and 3D Reconstruction: Benchmarking Seven Depth Estimation Models
Patil, S., Reddy, S., Asish, J., Prakash, R., & Prathap, B. R. (2024) · OITS International Conference on Information Technology (OCIT) · doi:10.1109/ocit65031.2024.00045
“Ranftl’s work on Dense Vision Transformers (DPT) [6] replaces traditional convolutional networks with vision transformers, providing improved global coherence and finer details for monocular depth estimation and semantic segmentation.”
Fan, W., Jiang, W., Fang, H., et al. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72089-5_53
PolarDepth: Polarization-Guided Monocular Depth for Visual Odometry
Rajyaguru, N., Wang, T., Tajne, A., He, B., Wu, J., Fermuller, C., Metzler, C. A., & Aloimonos, Y. (2026) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2026.3685941 · Cited by 1
“We build on DepthAnything-V2, keeping the DINOv2 encoder frozen [15] and fine-tuning only the DPT decoder [59].”
Huang, J., Guo, H., Lü, L., et al. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-05182-0_42 · Cited by 1
Desai, P. M., Kiran, G. U., Kumar, R. R., et al. (2026) · International Conference on Electronics and Sustainable Communication Systems (ICESC) · doi:10.1109/icesc70182.2026.11687484
Fan, X., Zhang, C., & Xu, Z. (2025) · International Conference on Information Science, Electrical and Automation Engineering (ISEAE) · doi:10.1109/iseae64934.2025.11042186
Deep in the Jungle: Towards Automating Chimpanzee Population Estimation
Raynes, T., Brookes, O., Haucke, T., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31933-3_10
Ma, J., Mingfeng, J., Fang, X., et al. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5134965
Pm-Neus:Progressive Sampling and Monocular Priors Enhanced Neural Surface ReconstructionPreprint
Yang, X., Jinhu, W., Wang, J., et al. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5385768
DeGraRec: 3D Deformable Object Reconstruction Using Graph Neural Networks and Depth Estimation
Loghin, M.-A., & Andreica, A. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-82021-2_11
Grasp Generation with Depth Estimation from Color Images
Nguyen, V.-T., Vu, V.-D., Hoang, N.-A., et al. (2024) · International Conference on Intelligent Information Technology (ICIIT) · doi:10.1145/3654522.3654575
“Tian et al. [3] integrated attention blocks into the decoder, while Transformer-based architectures gained traction [32, 42].”
RGB-D Salient Object Detection: A SurveyPreprint
Xiao, H., Li, J., Xiang, Y., Zhou, F., Liu, X., Yang, Z., & Tao, A. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7385363
Bayesian-Inspired Cross-Spectral Fusion Network for Robust Depth Estimation
Yan, J., Luo, W., Hu, Y., Zhang, C., Jing, M., & Zuo, L. (2025) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme59968.2025.11209550
“The depth head follows the decoder design of DPT [28].”
Choukhairi, O., Choukhairi, M., Choukri, A., & Fakhri, Y. (2026) · Journal of Imaging · doi:10.3390/jimaging12090435
“DPT [42] assembled ViT tokens from multiple depths into image-like feature maps.”
GFI-Net: Global Feature Interaction Network for Monocular Depth Estimation
Zhang, C., Xu, K., Ma, Y., et al. (2022) · International Conference on Pattern Recognition and Machine Learning (PRML) · doi:10.1109/prml56267.2022.9882253
Zhang, J., Zhan, X., Zhang, L., et al. (2026) · ACM SIGGRAPH Conference Papers (SIGGRAPH) · doi:10.1145/3799902.3811193
Research on Monocular Depth Estimation Method based on Multi-Level Attention and Feature Fusion
Wu, Z., Huang, H., Li, Q., et al. (2024) · IEEE Advanced Information Technology, Electronic and Automation Control Conference (IAEAC) · doi:10.1109/iaeac59436.2024.10503704
“Following the test protocol of [7~11], and without additional training data, our method improves the Absolute relative error by 1.1% , Log10 by 2.5%, and accuracy 𝛿 1 by 0.3% over PixelForm, which can be attributed to the proposed AFF module, which allows capturing long-range dependencies.”
MethodInfluentialInstance-Level Video Depth in Groups Beyond Occlusions
Liang, Y., Zhou, Y., Sun, Z., Xiang, T., Li, G., & He, S. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.00711
“DPT [42] introduced transformer-based architectures to enhance feature extraction, while synthetic datasets [36] and structure-from-motion techniques [30, 31] have been explored to mitigate data limitations.”
Shanto, H. M., Islam, M. T., Hasan, M. R., Pranto, S. S., Ahmed, T., Mahmud, R., & Tasnim, S. A. (2026) · AIUB Journal of Science and Engineering (AJSE) · doi:10.53799/zsh31s95
AlignGS: Aligning Geometry and Semantics for Robust Indoor Reconstruction from Sparse Views
Gao, Y., Zhong, H., Zhu, T., et al. (2025) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip67698.2025.11396845
“1) Depth Consistency Loss: We regularize the geometry by enforcing consistency between the rendered depth and priors from DINOv2 with a DPT head [23].”
XMDrone: Exploration of Indoor Drone Localization Using Exo-Monocular RGB Camera
Woga, M., Negash, R., Bekera, A. B., Alemu, N., Byambaa, D., Batdelger, M.-U., Batnasan, G., Abduljabbar, M., Alnajjar, F., Berengueres, J. O. L., et al. (2025) · IEEE International Conference on Application of Information and Communication Technologies (AICT) · doi:10.1109/aict67988.2025.11268730
“The base model architecture uses the DPT encoder-decoder [20].”
Depth-Guided Metric-Aware Temporal Consistency for Monocular Video Human Mesh Recovery
Cen, J., Mao, X., Yue, G., Zhou, W., Wang, R., Zhou, F. J., & Zhao, B. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11463978
FoundDP: Revisiting Weak Disparity Observability in Dual-Pixel Depth Estimation
He, F., Xu, H., Zhao, D., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37035-8_11
DNGaussian++: Improving Sparse-View Gaussian Radiance Fields With Depth Normalization
Li, J., Zhang, J., Yu, X., Bai, X. P., Zheng, J. Y., Ning, X., & Gu, L. (2026) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2026.3664307
“By default, we use DPT [65], [66] to predict monocular depth.”
Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens DistortionPreprint
Oskouei, S., Valla, M., Pedersen, A., Smistad, E., Dale, V. G., Høibø, M., Wahl, S. G. F., Haugum, M. D., Langø, T., Ramnefjell, M., et al. (2024) · Research Square · doi:10.21203/rs.3.rs-4601446/v1
GeoEdit: Geometry-Aware Object Editing via Dual-Branch Denoising
He, Y., Wang, J., Wang, X., Fong, M., Zhang, S., Xue, Y., Zheng, H.-T., & Ma, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37447-9_24 · Cited by 1
Real Time Steering Angle Estimation Using Monocular Depth Profiling and 1D Free Space Analysis
Ranjan, A., Diptanshu, K., Gnanendra Naidu, N., Maheshkumar, & Babu, B. S. (2025) · International Conference on Computation System and Information Technology for Sustainable Solutions (CSITSS) · doi:10.1109/csitss67709.2025.11294701
“Recent advances in deep learning have enabled increasingly accurate depth estimation from monocular RGB images [2], opening new possibilities for simplified autonomous navigation systems.”
A Robust Pipeline for Immersive E-Commerce Product Display Using Enhanced 3D Gaussian Splatting
Sudha, V. K., Sivakumar, V., K., et al. (2025) · doi:10.1109/icinvents64613.2025.11402301 · Cited by 1
“To overcome the inherent fragility of SfM, the pipeline augments this foundational step with a state-of-the-art AI-driven Monocular Depth Estimation (MDE) model, such as Depth Anything V2 [5].”
Progressive Learning for Semi-Supervised Ultrasound Image Segmentation
He, G., Xiang, X., Gong, Y., et al. (2025) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm66473.2025.11356878
“We compare against recent state-of-the-art methods using three encoders: ResNet34, ResNet50, and DINOV2-Small.”
FoundationGeo: Learning Spatial Pixel-Wise Fields for Monocular Metric Geometry
Liu, M., Lyu, X., Ren, T., Dai, P., Wu, X., Zhang, Z., Zhang, J., Lin, J., Shi, S., & Qi, X. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37092-1_20
Starting from Zero: A No-Pretraining Object DetectorsPreprint
Ye, X., Ma, T., & Xiao, B. (2024) · Research Square · doi:10.21203/rs.3.rs-4557206/v1
Adversarial Patches in 3D: A Stealth Threat to Monocular Depth Estimation
Pan, M. R., Dai, Y., Lai, J., & Xie, X. (2026) · Machine Intelligence Research · doi:10.1007/s11633-025-1617-6
Zhang, C., Fan, L., Cao, S., Wu, B., & Ye, J. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11128515
“The transition to vision transformers has notably enhanced performance [30], [31], [5].”
Transformers for prompt-level EMA non-response predictionPreprint
Nagesh, S., Moreno, A., Carpenter, S. M., Yap, J. R. T., Chatterjee, S., Lizotte, S. L., Wan, N., Kumar, S., Lam, C. Y., Wetter, D. W., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2111.01193 · Cited by 3
“Following their initial success in NLP [9, 31], transformers have proven effective in computer vision[11] [32] [17] and speech recognition [29], among other domains.”
InfluentialYiyu, 陈. C., Zicheng, 沈. S., & Yuanmu, 杨. Y. (2026) · Chinese Journal of Lasers · doi:10.3788/cjl260658
Mao, Y., Zhao, W., & Chen, L. (2026) · Agronomy · doi:10.3390/agronomy16100972
Self-Supervised Vision Transformers for Semantic SegmentationPreprint
Gu, X., Hu, Y., Wen, C., et al. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4785076
Zhou, W., Zou, Y., Gu, C., Dong, R., Xie, H., & Zhang, C. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37627-5_20
Mou, Z., Dai, Y., Cao, Z., Lv, Z., Wang, L. N., Shen, Y., & Fu, G. (2026) · Sensors · doi:10.3390/s26123662
POGS: Position-Optimized Gaussian Splatting for Reconstruction in Unevenly Illuminated Scenarios
Gao, C., Wang, C., Liu, J., et al. (2025) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc58881.2025.11343695
“…incorporates position supervision prior to color supervision, generating corresponding depth maps by leveraging a pre-trained monocular depth estimation network [4] and using these depth maps to constrain the positions of gaussian spheres, thus mitigating the impact of uneven illumination on 3DGS.”
Yaşar, H. A., & Ahrazoğlu, M. A. (2026) · International Conference on Recent Advances in Air and Space Technologies (RAST) · doi:10.1109/rast69551.2026.11672341
“…range and depth literature has repeatedly shown strong condition dependence at long range and under appearance variation [3], [5], so that even when average accuracy improves, single-frame metric inference remains vulnerable to scale ambiguity, viewpoint change, and evaluation leakage [6], [7].”
RGB Ensemble Strategies for Unsupervised Industrial Anomaly Detection on the AutoVI Dataset
López, S. V., Soria‐Olivas, E., & Sánchez-Montañés, M. A. (2026) · Electronics · doi:10.3390/electronics15102077 · Cited by 2
A Quality Metric for Semantically Transmitted Images in Machine-to-Machine CommunicationsPreprint
Gowrisetty, V. S. S., Lokumarambage, M., Samarathunga, P., et al. (2024) · Research Square · doi:10.21203/rs.3.rs-4616580/v1
Online Consistent Video Depth with Gaussian Mixture Representation
Liu, C., Eckart, B., & Kautz, J. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160785
“Performance comparisons We compare with state-of-theart single-frame based methods: MiDaS and DPT [41], [40].”
MethodSeeing Through Smoke: Surgical Desmoking for Improved Visual Perception
Lu, J., Jiang, F., Zhang, X., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38233-7_44
Elqattan, Y., Ismail, K., El-Kharashi, M. W., et al. (2026) · IEEE Access · doi:10.1109/access.2026.3696918
“More recent Metric Monocular Depth Estimation (MMDE) models seek to regress absolute depth directly by leveraging supervised metric annotations, camera-aware normalization, or large-scale mixed-dataset training [2], [3], [12], [16], [17], [18], [19].”
Image Segmentation Applied to Urban Surface and Aerial Constraints Analysis
Trani, M. L. A., & Madaschi, F. (2023) · Proceedings e report · doi:10.36253/979-12-215-0289-3.90
Pan, J., Xu, D., & Luo, Q. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.01517
“In the coarse stage, we employ a pre-trained monocular depth estimator [30] ˜ D to produce the depth map D i from the inpainted image I i .”
KineDepth: Utilizing Robot Kinematics for Online Metric Depth Estimation
Atar, S., Zhi, Y., Richter, F., & Yip, M. C. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11247739
“Monocular Metric Depth Estimation (MMDE) using neural networks was pioneered by transformer-based architectures [29].”
MaskingDepth: Masked Consistency Regularization for Semi-Supervised Monocular Depth Estimation
Baek, J., Kim, G., Park, S., An, H., Poggi, M., & Kim, S. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10801719
“Also, similar techniques were used to increase the robustness of Transformers for image-level or pixel-level classification [31], [32].”
InfluentialEnhanced depth completion for transparent objects via failure-aware RGB-D pretraining
Tao, T., Yang, J., Wu, W., et al. (2026) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2026.104922
Kwon, W., Cheon, S., & Hong, K. (2026) · ETRI Journal · doi:10.4218/etrij.2026-0062
PMPNet: Pixel Movement Prediction Network for Monocular Depth Estimation in Dynamic Scenes
Peng, K., Quarles, J. P., & Desai, K. (2022) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr56361.2022.9956096
“[9] proposed using the vision transformers as a backbone for monocular depth estimation.”
BackgroundMethod
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
