AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Understanding Image Representations by Measuring Their Equivariance and Equivalence

Karel Lenc, Andrea Vedaldi

IJCV 2018 · Architectures and efficiency › CNN design · approach: Steerable rotation-equivariant CNNs · 140 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 vertical-flip equivariant map at Conv1 instead of leaving features unchangedAlexNet · ILSVRC12 val (vertical flip)top-1 error: 0.75 -> 0.43Improved
add rotation 90 equivariant map at Conv5AlexNet · ILSVRC12 val (Rot 90)top-1 error: 0.75 -> 0.53Improved
learn equivariant mapping for horizontal flipAlexNet · ILSVRC12 val (horizontal flip)top-1 error: 0.44 -> 0.44No clear effect
learn equivariant mapping for scaling by 2^-1/2 at Conv5AlexNet · ILSVRC12 val (scale)top-1 error: 0.61 -> 0.50Improved
use task-oriented loss instead of feature-reconstruction forward selection from Conv3 onwards (ablation)CNN equivariant map · ILSVRC12 val (vertical flip)top-1 errorImproved
use forward selection instead of task-oriented loss up to Conv2 (ablation)CNN equivariant map · ILSVRC12 val (vertical flip)training samples neededImproved
use sparse regression (forward selection) instead of least squares (ablation)HOG equivariant map · HOG 5x5 cell arrayHellinger reconstruction errorImproved
add structured sparsity with neighbourhood m=1 instead of dense (m=inf) (ablation)forward selection k=5 · HOG regression cost 9x9learning time (s): 281.18 -> 5.91Improved
add structured sparsity with neighbourhood m=3 instead of dense (m=inf) (ablation)forward selection k=5 · HOG regression cost 5x5learning time (s): 12.21 -> 3.90Improved
stitch Places-trained first layers (Conv5) instead of ImageNet-trained into AlexNetFranken-CNN · ILSVRC12 valtop-1 error: 0.50 -> 0.65Hurt
use a learned stitching layer instead of identity map (ablation)Franken-CNN · ILSVRC12 valtop-1 error: >99% -> 0.43Improved
use equivariant regression with M_g at Conv5 instead of direct regressioncat head rotation regressor · VOC07 cat head rotationrotation error (degrees): 10.1 -> 13.4Hurt

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