Neural Nearest Neighbors Networks

NeurIPS 2018 Tobias PlötzStefan Roth

Non-local methods exploiting the self-similarity of natural signals have been well studied, for example in image analysis and restoration. Existing approaches, however, rely on k-nearest neighbors (KNN) matching in a fixed feature space... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Image Denoising BSD68 sigma25 N3Net PSNR 29.30 # 2
Image Denoising BSD68 sigma50 N3Net PSNR 26.39 # 4
Image Denoising BSD68 sigma70 N3Net PSNR 25.14 # 3
Image Super-Resolution Set5 - 4x upscaling N3Net PSNR 31.50 # 20
Image Denoising Urban100 sigma30 N3Net PSNR 30.19 # 2
Image Denoising Urban100 sigma50 N3Net PSNR 26.82 # 4
Image Denoising Urban100 sigma70 N3Net PSNR 25.15 # 2