AI Paper Atlas

Research map / Image generation / Image restoration

Neural Nearest Neighbors Networks

Tobias Plötz, Stefan Roth

NeurIPS 2018 · Image generation › Image restoration · approach: Priors for restoration networks · 141 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 end-to-end learned continuous nearest-neighbor processingDnCNN image denoising · Urban100, σ=25PSNR: 29.97 -> 30.19Improved
replace fixed-space KNN with a learned embedding and deterministic continuous neighbors (ablation)DnCNN image denoising · Urban100, σ=25PSNR: 30.08 -> 30.19Improved
use denoised rather than noisy patches for hard KNN matching (ablation)DnCNN image denoising · Urban100, σ=25PSNR: 30.07 -> 30.08No clear effect
use stochastic Concrete relaxation instead of deterministic continuous neighbors (ablation)N³Net image denoising · Urban100, σ=25PSNR: 30.19 -> 29.97Hurt
reduce the architecture from 3 × DnCNN blocks to 2 × DnCNN blocks (ablation)N³Net image denoising · Urban100, σ=25PSNR: 30.19 -> 29.99Hurt
increase the selected-neighbor count up to four (ablation)N³Net image denoising · Urban100, σ=25 and σ=50PSNR: k=1: 30.17; k=4: 30.27; k=7: 30.19; k=1: 26.76; k=4: 26.86; k=7: 26.82Mixed
augment DnCNN with an N³ blockDnCNN image denoising · Set12, BSD68, and Urban100 at σ=25, 50, and 70PSNR: Set12: 30.44 -> 30.55 (σ=25); 27.19 -> 27.43 (σ=50); 25.56 -> 25.90 (σ=70); BSD68: 29.23 -> 29.30 (σ=25); 26.23 -> 26.39Improved
augment VDSR with two N³ blocksVDSR super-resolution · Set5 ×2, ×3, and ×4PSNR: 37.53 -> 37.57; 33.66 -> 33.84; 31.35 -> 31.50Improved
insert an N³ block after the sixth CNNet blockCNNet correspondence classification · St. Peter / St. Peter, 5°MAP: 0.271 -> 0.316Improved
insert an N³ block after the sixth CNNet blockCNNet correspondence classification · St. Peter / Reichstag, 5°MAP: 0.173 -> 0.231Improved
insert an N³ block after the sixth CNNet blockCNNet correspondence classification · Brown / Brown, 5°MAP: 0.236 -> 0.293Improved
apply a post-hoc NN3D non-local stepDnCNN · BSD68, σ=50PSNR: -0.04 dB compared to DnCNNHurt

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