AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Denoising Vision Transformers

Jiawei Yang, Katie Z Luo, Jiefeng Li, Congyue Deng, Leonidas Guibas, Dilip Krishnan, Kilian Weinberger, Yonglong Tian, Yue Wang

ECCV 2024 · Architectures and efficiency › Attention and transformers · approach: Hybrid and interpretable ViT designs · 51 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add DVT denoiser to pre-trained ViTsDINOv2 ViT-B · VOC2012 segmentationmIoU: 83.60 -> 84.84Improved
add DVT to DINOv2-regDINOv2-reg · VOC2012 and ADE20k segmentationmIoU: +0.86 in VOC2012 and +1.12 in ADE20kImproved
add DVT denoiserDeiT-III, DINO, EVA02, CLIP, DINOv2 · NYUv2 depthdelta_1: DeiT-III 72.16 -> 71.36; DINO abs rel 0.1701 -> 0.1731Mixed
add DVT denoiserDINOv2-reg · VOC2007 object discovery (LOST)corloc: 38.0 -> 56.1Improved
add DVT denoisersix ViTs · VOC detectionmAP: DeiT-III 75.8 -> 77.0Improved
add artifact field G and residual term to the semantics field (ablation)per-image denoising · VOC12 val KNN segmentationmIoU: 67.81 -> 70.94Improved
add residual term on top of F + G (ablation)per-image denoising · VOC12 val KNN segmentationmIoU: 70.82 -> 70.94No clear effect
use transformer block denoiser instead of conv1x1 (ablation)generalizable denoiser · VOC2012 valmIoU: 82.15 -> 84.84Improved
add learnable positional embeddings to the denoiser (ablation)single Transformer block denoiser · VOC2012 valmIoU: 84.81 -> 84.84No clear effect
use more views in per-image neural field denoising (ablation)DVT stage one · dense prediction taskstask performanceImproved
train denoiser on aggregated-view features only (F-only) instead of full decomposition (ablation)DVT · dense prediction taskstask performanceHurt
train denoiser with only 100 to 1000 denoised samples (ablation)DVT denoiser · dense prediction taskstask performanceImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy