Scale-invariant scale-channel networks: deep networks that generalise to previously unseen scales

DOI10.1007/s10851-022-01082-2OpenAlexW3171742925MaRDI QIDQ2163348

Tony Lindeberg, Ylva Jansson

Publication date: 10 August 2022

Published in: Journal of Mathematical Imaging and Vision (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2106.06418

zbMATH Keywords

scale invariancescale spacescale covarianceconvolutional neural networksdeep learningscale generalisationinvariant neural networks

Mathematics Subject Classification ID

Computer science (68-XX) Information and communication theory, circuits (94-XX)

Uses Software

• ASIFT
• DeepLab
• SIFT
• OverFeat
• CIFAR
• DeepFool
• ScatNet
• SURF
• ESPNet
• Faster R-CNN
• RotEqNet

Cites Work

• Unnamed Item
• Linear scale-space has first been proposed in Japan
• A computational theory of visual receptive fields
• The structure of images
• Generalized Gaussian scale-space axiomatics comprising linear scale-space, affine scale-space and spatio-temporal scale-space
• On the axioms of scale space theory
• Provably scale-covariant continuous hierarchical networks based on scale-normalized differential expressions coupled in cascade
• Image matching using generalized scale-space interest points
• Scale selection properties of generalized scale-space interest point detectors
• ASIFT: A New Framework for Fully Affine Invariant Image Comparison
• Learning Rotation-Invariant and Fisher Discriminative Convolutional Neural Networks for Object Detection
• Scale-covariant and scale-invariant Gaussian derivative networks
• Scale-covariant and scale-invariant Gaussian derivative networks

This page was built for publication: Scale-invariant scale-channel networks: deep networks that generalise to previously unseen scales