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 2,001–2,100 of 2,248
Enhancing Local Information and Long-Distance Dependencies for Finer Monocular Depth Estimation
Huo, Z., Zhao, L., & Chen, Y. (2023) · International Congress on Image and Signal Processing (CISP) · doi:10.1109/cisp-bmei60920.2023.10373349
“Monocular depth estimation has been extensively studied as a fundamental problem in computer vision [12], [13], [9], [5] and [14].”
BackgroundMethodLiu, Z., Xue, J., Wu, P., Xie, Z., & Wang, X. (2026) · Journal of King Saud University - Computer and Information Sciences · doi:10.1007/s44443-026-01175-7
“Monocular methods such as MiDaS, DPT, and Depth Anything provide geometric priors without dedicated depth cameras (Ranftl et al. 2022, 2021; Yang et al. 2024a, b).”
Soumya, A., Cenkeramaddi, L. R., & Krishna Mohan, C. (2026) · Applied Intelligence · doi:10.1007/s10489-026-07190-5
“In the context of object detection, DETR [21] avoids the need for hand-crafted anchors by using bipartite matching for detection, and dense prediction transformers [22] extend transformers for use in pixel-level dense prediction tasks like depth estimation and semantic segmentation.”
CORE: Co-planarity Regularized Monocular Geometry Estimation with Weak Supervision
Li, Y., Wang, K., Li, H., et al. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00808
“As the rising of Transformer architectures [10, 28], the more recent approaches [5, 35, 43, 26, 1] further improve the depth estimation performance by better global perception.”
BackgroundAttention-Guided Shared Hybrid Network for Enhanced Land Cover Segmentation
Jiang, Y., Shi, L., & Fan, X. (2024) · IEEE Access · doi:10.1109/access.2024.3524454
Xie, P., Zhou, F., & Shao, S. (2026) · Electronics · doi:10.3390/electronics15184234
PAAP: Performer-Aware Automatic Panning System
Lee, K., & Kim, S. (2025) · ACM SIGGRAPH Conference (SIGGRAPH) · doi:10.1145/3721250.3742989
“Our system leverages OC-SORT [Cao et al. 2023] to track performers and MiDaS [Ranftl et al. 2021, 2022] to extract depth feature maps which are essential for spatial information algorithm.”
Geometric Distillation from Rectified Stereo: Leveraging Epipolar Cues for Monocular Depth
Kim, J.-H., & Liu, X. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37362-5_34
3D scene reconstruction from images
Ramamonjisoa, M. (2022) · theses.fr (ABES) · doi:10.70675/b55ee9f9z4bc1z401fz800ez119f7f7ceaa3
An Enhanced Keyframe Extraction Technique based on Multi-Object Tracking and Re-Identification
Lee, S., & Min, B. (2026) · Journal of Broadcast Engineering · doi:10.5909/jbe.2026.31.3.373
MSD-CRFS: Multi-Scale Dual Aggregation Conditional Random Fields for Monocular Depth Estimation
Zhang, X., Wei, J., Moteki, A., et al. (2024) · Proceedings - International Conference on Image Processing · doi:10.1109/icip51287.2024.10647895
Vulnerability of Privacy-Preserving Visual Localization Methods Against Diffusion-Based Attacks
Pietrantoni, M., Sattler, T., & Csurka, G. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37095-2_27
DRMip-Gaussian: Unaliasing Sparse-View 3D Gaussian Splatting with Depth Regularization
Lin, T., Bai, Z., Xu, M., et al. (2025) · International Conference on Video, Signal and Image Processing (VSIP) · doi:10.1145/3784713.3784726
Terzi, R., & Demi̇rezen, M. U. (2022) · Journal of Ankara University Faculty of Medicine · doi:10.4274/atfm.galenos.2022.97830
DeFT-LoRA-C: Depth Estimation with Foundation Model Tuning Using Low-Rank Adaptation for ColonoscopyPreprint
Carlini, L., Magro, M., Lena, C., Massimi, D., Rizkala, T., Islam, M., Hassan, C., & Momi, E. D. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5325702
Vision-Based Pose Measurement System for Intermodal Containers
Zheng, L., He, M., Zhang, Y., et al. (2026) · International Conference on Intelligent Environments (IE) · doi:10.1109/ie69249.2026.11538981
“To address complex scenarios, Transformers [6] and state space models (SSMs) like Mamba [7, 8] have emerged, substantially improving global modeling capabilities.”
Physics-Informed Input and Decomposed SSIM for Accurate Single-Shot 3D Reconstruction
Nguyen, K., Dinh, K. T., TonThat, L., et al. (2026) · International Conference on Communications and Electronics (ICCE) · doi:10.1109/icce70609.2026.11658467
Multi-expert Representation Learning with Dynamic Routing for Brain Captioning
Guo, Z., Xia, W., Cao, M., Zhou, C., & Öztireli, C. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38239-9_35
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
Metrological Assessment of Rail Clip Fasteners via Monocular 3D Reconstruction
Gallo, V., Shallari, I., Laino, V., et al. (2026) · doi:10.1109/ai4im69129.2026.11558207
“Transformer-based dense prediction (DPT) improved coherence and detail [2], [3], while Depth Anything emphasized robustness “in the wild,” again prioritizing broad generalization over strict metric calibration unless metric supervision is added [3], [4].”
Getting Started with Semantic Segmentation in PyTorch Using SMP
Mari, J. F., Silva, L. H. F. P., Escarpinati, M. C., et al. (2025) · SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI) · doi:10.1109/sibgrapi67909.2025.11223385
Sherif, M. E., & Mansour, M. N. M. (2026) · International Conference on Computational Intelligence Approaches and Applications (ICCIAA) · doi:10.1109/icciaa68481.2026.11543857
“Automated depth estimation and instance segmentation have become especially important because they provide spatial and semantic control signals for AI-assisted VFX work [5], [6].”
MDEConvFormer: estimating monocular depth as soft regression based on convolutional transformer
Su, W., Ye, H., Zhang, H., et al. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18290-0
“The performances of our proposed method are slightly lower than [12, 13], and [27].”
InfluentialOmniX: Any-View and Any-Time 4D Reconstruction via Feed-Forward Trajectory Fields
Jiang, Y., Wang, T., Wang, Z., Cao, C., Wu, J., Luo, W., Hu, W., Gao, J., & Guo, C. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37235-2_19 · Cited by 3
Finding NeMO: A Geometry-Aware Representation of Template Views for Few-Shot Perception
Jung, S., Klüpfel, L., Triebel, R., & Durner, M. (2026) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv69130.2026.00039
“As ViT we use DINOv2 [45] and DPT [48] as regression head.”
Deep Learning Based Face liveness detection with a Lightweight CNN and depth detection using MiDaS
Das, N., Singh, M. P., Biswas, P., et al. (2026) · doi:10.1109/cipher70417.2026.11523927
“As a result, recent work has investigated monocular depth estimation techniques, such as MiDaS, which enable depth-based face liveness detection using standard RGB images without additional sensors [6].”
Design of a Background Extraction Platform with Semantic Elements
Fernández-Badilla, J., Chavarria-Zamora, L., Barboza-Artavia, L., et al. (2025) · IEEE International Conference on BioInspired Processing (BIP) · doi:10.1109/bip68491.2025.11489147
“Monocular depth estimation has advanced from self-supervised stereo-style training (e.g., Monodepth2) to dataset-agnostic transformer-driven estimators (e.g., MiDaS/DPT) [5], [15], [16].”
GCDepth: Geometry-Conditioned Autoregressive Monocular Depth Estimation
Wang, K., Ji, Y., Shen, T., et al. (2026) · Youth Academic Annual Conference of Chinese Association of Automation (YAC) · doi:10.1109/yac71005.2026.11615632
Vision-Driven 2D Supervised Fine-Tuning Framework for Bird’s Eye View Perception
He, L., Wang, Q., Sun, H., Xu, Q. C., Gao, B., Li, S. E., Wang, J., & Li, K. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11246559 · Cited by 1
“Early works like WS3D [25] utilized 2D labels and point cloud data with limited accurate 3D annotations to automatically generate 3D labels for training, thereby effectively reducing the cost of object detection and 3D data annotation.”
Robust visual semantic perception for flexible grinding of complex welds
Wu, J., Qiu, W., Huang, J., & Chen, H. (2026) · Biomimetic Intelligence and Robotics · doi:10.1016/j.birob.2026.100288
Optidepth:Edge-AI Optimized Monocular Depth Estimation for Real-Time Heavy Machinery Safety
Jeyashree, A., S, S. K. C., Jayachandran, M., et al. (2025) · International Conference on Intelligent Technologies for Sustainable Electric and Communications Systems (iTech SECOM) · doi:10.1109/itechsecom64750.2025.11307603
A Two-Stage Method for Detection and Distance Perception of Traffic Lights
Sarker, S., Pramanik, A., & Sarkar, S. (2026) · Lecture notes in civil engineering · doi:10.1007/978-981-96-9642-0_36
Depth-Aware Spatial Hierarchy Learning for Robust Medical Image Segmentation
周, 世. (2026) · Software Engineering and Applications · doi:10.12677/sea.2026.153046
Gassenn, A. E., Machado, M. C., & dos Santos, E. M. (2024) · doi:10.5753/sbcas.2024.2191
Single Image Depth Estimation Based on CLIP Language Prior Injection
Yu, S. C., Miao, J., & Wang, J. (2026) · Journal of Multimedia Information System · doi:10.33851/jmis.2026.13.2.43
Zuo, Y., Wen, H., Subramanian, V., Chen, P., Kayan, K., Bijelic, M., Heide, F., & Deng, J. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37422-6_1 · Cited by 1
Fan, D., & Liu, S. (2024) · Lecture notes in computer science · doi:10.1007/978-981-97-5612-4_31
Guo, Y., Zhou, C., Xiao, C., Yuan, H., & Li, Q. (2026) · IEEE Sensors Journal · doi:10.1109/jsen.2026.3668036
“The selected baseline methods include dense prediction transformer (DPT) [43] and MonoDepth2 [44], both of which are representative models widely used in monocular depth estimation.”
Pose-Based Pedestrian Localization: A Comparative Study of Stereo and Monocular Depth Estimation
Gamal, M., Badawy, A., Elnabawy, I., et al. (2025) · International Japan-Africa Conference on Electronics, Communications, and Computations (JAC-ECC) · doi:10.1109/jac-ecc67970.2025.11417501
“Monocular depth models, including DPT [8], Depth Anything [9] and LiteMono [11], offer faster inference but exhibit 1–2.1 m MPJPE depending on architecture and input resolution.”
InfluentialOmniPoint: Universal Monocular Metric Pointcloud from Any Camera
Ye, B., Pollefeys, M., Yang, M., & Kundu, A. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37602-2_32
Remote Sensing Object Detection in Disturbed Environment Based on Improved RTMDet Network
Yang, J., Liu, M., Dong, S., Zheng, R., & Wei, P. (2024) · Chinese Control Conference (CCC) · doi:10.23919/ccc63176.2024.10662723
“We propose a module based on RTMDet with a large kernel size to augment the receptive field [7, 10, 16, 23] and capture a broad context.”
MethodA Review of Vision-Based Depth Estimation: Current Methods and Future Directions
Zhang, X., Song, Y., & Lee, D. (2025) · IEEE International Conference on Mobile Ad-Hoc and Smart Systems (MASS) · doi:10.1109/mass66014.2025.00093
“• DPT (Vision Transformer for Dense Prediction) [21] adopts transformer-based backbones to incorporate long-range dependencies.”
A Survey of Monocular Depth Estimation based on Deep Learning
Song, Z., & Jianxing, W. (2024) · Frontiers in Computing and Intelligent Systems · doi:10.54097/7iuoscdg
HM-NVS: Hierarchical Multi-Modal Novel View Synthesis\\with Uncertainty-Aware Progressive Refinement
Chang, J., Zhao, H., & Zhang, L. (2026) · International Conference on Multimedia Retrieval · doi:10.1145/3805622.3810569
GeoNVS: Geometry Grounded Video Diffusion for Novel View Synthesis
Kang, M., Shin, I., Lee, T., Kim, M., Kweon, I. S., & Yoon, K. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37013-6_30
ArtArchitect: Adaptive Depth Correction and Outpainting for Realistic World Generation
Zhang, Y., Chan, H. C. B., Duan, R., Yu, H., Wang, X., & Lin, H. (2025) · IEEE Annual International Computer Software and Applications Conference (COMPSAC) · doi:10.1109/compsac65507.2025.00070
Emergent Vision Technology: 3D Human Pose Estimation for Single-Pixel Imaging (SPI)Preprint
Quero, C. A. O., & Martínez-Carranza, J. (2024) · Research Square · doi:10.21203/rs.3.rs-4837829/v1
RenHaze: A Coarse-to-Fine Rendering Framework for Improving Robustness to Haze
Le, T.-H., Huang, S.-C., Hoang, Q.-V., et al. (2025) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3770745
Decoding Visual Acceptability in Ecological Restoration: A Multi-Modal Explainable AI Approach
Kang, Y., Kim, E. J., & Son, S. W. (2026) · Land · doi:10.3390/land15091711
Navigating the wild: Pareto-optimal visual decision-making in image space
Pushp, D., Chen, W., Chen, Z., Luo, C., Gregory, J. M., & Liu, L. (2025) · The International Journal of Robotics Research · doi:10.1177/02783649251399686
“Recently, Transformer-based models (Vaswani et al. 2017; Dosovitskiy et al. 2020; Ranftl et al. 2021) have gained prominence in semantic segmentation.”
SoMADepth: Fine-Grained Geometry via Generative Priors for Monocular Depth Estimation
Daosheng, W. A. N. G., Ming, Y., & Guo, H. (2026) · Cross Strait Radio Science and Wireless Technology Conference (CSRSWTC) · doi:10.1109/csrswtc71154.2026.11669207
“SoMADepth uses a DINOv3-ViT-L/16 encoder [9] and a UniDepth-style camera-aware decoder [3], [7], with the diffusion prior following Pixel-Perfect Depth [17].”
Zami, A., Sobhan, S., Ahmed, M., et al. (2026) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2026.110162
Santhanan, S., Anirudh, A., Adhavan, R., & Dhiraj, D., V (2026) · International Conference on Electronics and Renewable Systems (ICEARS) · doi:10.1109/icears67481.2026.11416655
Tsukanov, A. I., & Chekanin, V. A. (2025) · International Russian Automation Conference (RusAutoCon) · doi:10.1109/rusautocon65989.2025.11177401
“We show that Depth Anything V2, pretrained on a massive and diverse depth dataset, provides substantially stronger foundational features compared to models like DPT [2] with less extensive training or models trained from scratch.”
InfluentialMutee, A. F., Al-Hussainy, A. F., Ibrahim, N. A., et al. (2025) · Abdominal Radiology · doi:10.1007/s00261-025-05295-y
Multi-scale Feature Distribution Prediction Decoder for Lightweight Monocular Depth Estimation
Ma, X., Ji, Y., Song, C., & Zhong, S. (2024) · International Conference on New Trends in Computational Intelligence (NTCI) · doi:10.1109/ntci64025.2024.10776079
Chornenkyi, V. Y., & Kazymyra, I. (2024) · Ukrainian Journal of Information Technology · doi:10.23939/ujit2024.02.074
EDADepth: Enhanced Data Augmentation for Monocular Depth Estimation
Khanal, N., & Sheshappanavar, S. V. (2024) · International Conference on Machine Learning and Applications · doi:10.1109/icmla61862.2024.00090 · Cited by 3
Complete information extraction for monocular depth estimation using a dual framework
Li, B., Zhou, D., Gao, X., et al. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2025.112337
Transformer network for stereo matching of weak texture objects
Jia, D., Peng, C., Wu, S., et al. (2024) · Journal of Image and Graphics · doi:10.11834/jig.230575
AURA-Depth: Attention-Based Uncertainty Reduction and Feature Aggregation Depth Network
Na, Y., Woo, S., Lee, S., & Kang, B. J. (2025) · IEEE Access · doi:10.1109/access.2025.3556062
“With the advent of deep learning, Convolutional Neural Networks (CNNs) and Transformer-based [17] architectures have significantly improved feature representation and global context understanding [18], [19], [20], [21], [23], [24], [25], [26], [27], [28].”
Does Image Deraining Improve Downstream Perception Tasks in Autonomous Driving?
Bitwire, G. A., Kim, J. H., & Han, D. S. (2026) · International Conference on Ubiquitous and Future Networks · doi:10.1109/icufn69619.2026.11628692
“Mi-DaS [21], [22] and DPT [23] achieve strong cross-dataset generalization, while methods such as MonoDTR [2], Mon-oDETR [3], MonoDGAE [4], and EDATRAF [5] demonstrate competitive 3D perception using monocular inputs.”
Semantic-Aware Autonomous Vehicle Navigation with Causal Transformers and Q-Learning
Pandya, M., Siddalingaswamy, P. C., & Singh, S. (2025) · ACM Journal on Autonomous Transportation Systems · doi:10.1145/3749992
“We benchmarked our depth component against approaches like Monodepth2 [18], Pre-trained Midas model [50], Unsupervised MDE with semantic aware representation [7] along with two other approaches of [3] and [66] which utilize semantic segmentation priors for understanding scene depths.”
InfluentialWang, Y., Yuan, X., Yue, Z., Kong, Y., & Lin, S. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.116471
Learn to Rank: Visual Attribution by Learning Importance Ranking
Schinagl, D., Fruhwirth-Reisinger, C., Prutsch, A., Schulter, S., & Possegger, H. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37592-6_20
A Traffic Accident Prediction Model Based on Spatio-Temporal Depth Map and Transformer
Zhou, C., Wang, H., Yang, J., et al. (2025) · International Conference on Computer Supported Cooperative Work in Design (CSCWD) · doi:10.1109/cscwd64889.2025.11033325
Attention-CARU With Texture-Temporal Network for Video Depth Estimation
Im, S., & Chan, K. (2025) · IEEE Access · doi:10.1109/access.2025.3579429
ViewSplat: View-Adaptive 3D Gaussian Splatting for Feed-Forward Synthesis
Jeong, M., Min, S., Lee, S., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37624-4_14
Employing Vision Transformer for Monocular Depth Estimation
Chaudhary, A., Tripathi, D. B., Mishra, H., Pathak, P., & Dixit, A. (2025) · International Conference on Computational Intelligence, Communication Technology and Networking (CICTN) · doi:10.1109/cictn64563.2025.10932519
Forkuo, G. O. (2026) · Research Square · doi:10.21203/rs.3.rs-8908445/v1
Depth-Copy-Paste: Multimodal and Depth-Aware Compositing for Robust Face Detection
Guo, Q., Jiang, X., Wang, Y., et al. (2026) · International Conference on Automatic Face and Gesture Recognition · doi:10.1109/fg67764.2026.11556984
“More recently, transformer-based models and large-scale training pipelines, e.g., DPT [18], MiDaS [19], and Depth-Anything [5] have significantly improved robustness and cross domain generalization through massive heterogeneous datasets and strong global-reasoning backbones.”
Pseudo Progressive Global to Local Transformer Network for Segmentation Refinement
Yan, H., Wu, G., & Ding, J. (2025) · IEEE Transactions on Emerging Topics in Computational Intelligence · doi:10.1109/tetci.2025.3604819
“Notably, Ranftl et al. [32] construct multiple vision transformers that progressively integrate their multi-scale features to generate dense predictions.”
Reliability-Guided Consistency Learning for Robust Semi-Supervised Semantic Segmentation
Xiong, Q., Yan, G., Li, X., et al. (2026) · IEEE Transactions on Consumer Electronics · doi:10.1109/tce.2026.3664484
“We build a lightweight segmentation framework based on DPT [42] and use DINOv2 pretrained encoders, including DINOv2-small and DINOv2-base [29].”
MambaDS U-Net: A multi-scale mamba and dynamic upsampling enhanced U-Net for glomeruli segmentation
Li, G., Deng, J., Zhang, Y., et al. (2026) · Signal Image and Video Processing · doi:10.1007/s11760-026-05311-8
“Under the defined KPI dataset, this paper conducts comparative experiments on mainstream segmentation networks, including DPT [33], Attention U-Net [34], U-Net [8], CPCANet [35], MA-Net [36], U-Net++ [9], UPerNet [37], Swin-UNet [22], TransUNet [21], and SegFormer [38].”
Do, J.-S., & Jung, H. (2026) · Sensors · doi:10.3390/s26134192
“Following the scale-and-shift-invariant alignment convention used in relative-depth MDE evaluation [7,8], the pair minimizes the squared error against the ground-https truth target over the valid-pixel set M f defined by Equation (2) in the SoccerNet-Depth normalized target domain: The aligned…”
Aggregating Cross-Domain Knowledge via Learnable Tokens for Multi-teacher Distillation
Ran, W., Zhang, W., Pang, S., Liu, J., Zhang, X., Yan, Y., & Ma, C. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37138-6_3
MGS-SLAM: Monocular 3D Gaussian Splatting SLAM with Significance-Guided Pruning
Cao, T., Huang, S.-S., Guo, K., Sha, H., Li, D., & Liu, Y. (2026) · Computational Visual Media · doi:10.26599/cvm.2026.9450555
Multi-Scale and Multi-Modal Selective Fusion for RGB-D Video Salient Object Detection
Liu, Y., He, J., Fu, K., et al. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11461712
Efficient depth from defocus imaging for environmental perception
Huang, Y., Li, Y., Tian, J., et al. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.129217
3D Reconstruction of Colonic Polyps: A Methodology for Improved Visualization and Analysis
Ahmed, I., Saha, P., Hamim, M. S. I. K., et al. (2025) · International Conference on Electrical, Computer and Communication Engineering · doi:10.1109/ecce64574.2025.11013942
“Segmented photos are used to generate depth maps using the Intel DPT-large model [7], enabling accurate depth estimate from two-dimensional images.”
InfluentialSelf-Supervised Learning for Context-Independent DfD Network using Multi-View Rank Supervision
Mishima, N., Seki, A., & Hiura, S. (2023) · Proceedings - International Conference on Image Processing · doi:10.1109/icip49359.2023.10222834
“1 [4] 0.135 0.744 MiDaS v3 [5] 0.114 0.782 Phone MuRaS (ours) Defocus 0.044 0.925 Phys.”
BackgroundMethodInfluentialUncertainty-Gated Dual-path Probabilistic Depth Fusion for Robot Manipulation
Zhou, T., Ma, B., Satirapod, C., Gu, C., Pan, G., Yang, S., Yu, W., & Wang, S. (2026) · Youth Academic Annual Conference of Chinese Association of Automation (YAC) · doi:10.1109/yac71005.2026.11616164
“These approaches typically construct spatial representations from depth estimation [12]–[14], point clouds [15], [16], or multi-view geometry [17], [18], and integrate them into manipulation policies to improve geometric reasoning.”
Efficient Monocular Depth Estimation Via Single-Step Latent Diffusion Models
Huo, Z., & Wang, Z. (2025) · International Conference on Multimedia Retrieval · doi:10.1145/3731715.3733319
A Comparative Study on the Accuracy and Constraints of Pose-Free Gaussian Splatting Methods
Kawka, M. S., & Γραμματικόπουλος, Λ. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-29558-3_5
Optimizing Depth and Ego-Motion Estimation with Swin Transformer: A Comprehensive Study
Hossain, M. S., Hasan, S. M. M., Mamun, M. A., Srizon, A. Y., Faruk, M. F., & Hossain, M. R. (2024) · IEEE International Conference on Power, Electrical, Electronics and Industrial Applications (PEEIACON) · doi:10.1109/peeiacon63629.2024.10800482
“Meanwhile, DeiT [7] introduces a distillation to-ken to lessen the requirement for huge datasets, while DPT [8] stacks ViTs for hierarchical features but faces size concerns.”
Cao, R., Li, W., Tai, S., Chen, Z., Zhao, Y., Wu, Y., Zhao, S., Zhang, Y., & Xu, Y. D. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.116638
2D vision-based with monocular depth estimation for pose estimation
Charoenwuttikajorn, P., Ekkachai, K., Horanont, T., et al. (2025) · International Conference on Image Processing and Machine Vision (IPMV) · doi:10.1117/12.3062820
“R. Ranftl et al. [14] described how DPT improved performance on monocular depth estimation, improving it by up to 28% compared to former models.”
FeDepthX: A Federated Learning Depth eXperiment
Fraux, D., Druart, A., & Marat, S. (2025) · International Conference on Image Processing Theory, Tools and Applications (IPTA) · doi:10.1109/ipta66025.2025.11222036
EReFormer-DSC: Token-Compatible Dynamic Sampling for Event-based Monocular Depth Estimation
Bai, S., Li, Z., & Zhao, J. (2026) · doi:10.1145/3838457.3838512
“For conventional frame-based depth estimation, approaches including DPT [4], Swin-Depth [10], and DepthFormer [7] have validated that Transformer and CNN-Transformer hybrid structures can produce more consistent and spatially precise results than pure CNN models.”
Comparative Evaluation of Monocular Depth Estimation Models for Image Based Vehicle Dent Assessment
Thimbiripalage, D. D., & Sammani, M. H. K. (2026) · International Conference on Circuit, Systems and Communication (ICCSC) · doi:10.1109/iccsc71566.2026.11650139
“Transformer (DPT-Large), which used a Vision Transformer (ViT) backbone in place of traditional CNN-based decoders for dense depth prediction [7].”
Yoon, H., & Kim, H. K. (2025) · International Conference on Control, Automation and Systems (ICCAS) · doi:10.23919/iccas66577.2025.11301238
“MiDaS [8], a robust monocular depth estimation framework trained across multiple diverse datasets to enable strong zero-shot cross-dataset transfer.”
Robust Multi-person Reconstruction in Right Spatial Arrangement from In-the-Wild Scenes
Wang, W., Zhang, R., Huang, H., et al. (2023) · Lecture notes in electrical engineering · doi:10.1007/978-981-19-9968-0_49
Boosting Omnidirectional Stereo Matching with a Pre-trained Depth Foundation Model
Endres, J., Hahn, O., Corbière, C., Schaub-Meyer, S., Roth, S., & Alahi, A. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11246233
“This foundation model consists of a ViT encoder [18] and a Dense Prediction Transformer (DPT) decoder [58] to predict dense relative depth from the ViT patch encodings.”
Monocular Depth Estimation and 3D Point Cloud Reconstruction Using U-Net in Indoor Environments
Zhang, J.-L., & HIRAOKA, T. (2026) · Journal of the Japan society of photogrammetry and remote sensing · doi:10.4287/jsprs.65.182
DRG-LiteStar-YOLO: Reliability-Guided Pseudo-Depth Fusion for Dairy Goat Detection in Complex Barns
Zhang, Y., Wang, K., Yue, Y., et al. (2026) · Animals · doi:10.3390/ani16182840
Bi-projection Fusion for 360 Depth Estimation Based on Coordinate Attention
Li, L., He, C., Yu, Y., et al. (2026) · IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) · doi:10.1109/mipr70517.2026.00045
“In perspective imaging, deep learning methods [1]– [3] and Transformer-based architectures [4], [5] have achieved remarkable success.”
Beyond Random Sampling: Distribution-Aware Alignment for Semi-supervised Medical Image Segmentation
Yan, W., Qian, Y., Dong, Y. H., & Yang, M. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37132-4_21 · Cited by 1
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
