AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Learning on the Edge: Investigating Boundary Filters in CNNs

Carlo Innamorati, Tobias Ritschel, Tim Weyrich, Niloy J. Mitra

IJCV 2019 · Architectures and efficiency › CNN design · approach: Filter design and trained-filter analysis · 65 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
use explicit boundary filters instead of zero paddingU-net-like CNN · Gauss filteringDSSIM: .0022 -> .0016Improved
use explicit boundary filters instead of reflect paddingU-net-like CNN · Gauss filteringDSSIM: .0018 -> .0016Improved
use explicit boundary filters instead of zero paddingencoder-decoder · De-noise/BayerPSNR: 31.50 -> 31.94Improved
use explicit boundary filters instead of zero paddingencoder-decoder · ColorizationDSSIM: .1604 -> .1577Improved
use explicit boundary filters instead of zero paddingencoder-decoder · Disparitypixel error: 1.511 -> 1.403Improved
use explicit boundary filters instead of zero paddingencoder-decoder · Scene flowpixel error: 1.183 -> 1.096Improved
use explicit boundary filters instead of reflect paddingencoder-decoder · Scene flowMSE ratio: 56% -> 100%Improved
use explicit boundary filters instead of zero/reflect while varying network depth and width (ablation)simple Gaussian blur CNN · Gaussian blur learningMSEImproved
replace nine-convolution implementation with domain decomposition (ablation)explicit boundary filters · Boundary overheadruntime: 22.5 -> 2.5 msImproved
replace per-pixel conditional kernel select with domain decomposition (ablation)explicit boundary filters · Boundary overheadruntime: 11.2 -> 2.5 msImproved
add explicit boundary filters in Keras training (ablation)encoder-decoder · Training time per epochperformance loss: 0.2%No clear effect

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