Compressing the Input for CNNs with the First-Order Scattering Transform - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2018

Compressing the Input for CNNs with the First-Order Scattering Transform

Abstract

We study the first-order scattering transform as a candidate for reducing the signal processed by a convolutional neural network (CNN). We study this transformation and show theoretical and empirical evidence that in the case of natural images and sufficiently small translation invariance, this transform preserves most of the signal information needed for classification while substantially reducing the spatial resolution and total signal size. We show that cascading a CNN with this representation performs on par with ImageNet classification models commonly used in downstream tasks such as the ResNet-50. We subsequently apply our trained hybrid ImageNet model as a base model on a detection system, which has typically larger image inputs. On Pascal VOC and COCO detection tasks we deliver substantial improvements in the inference speed and training memory consumption compared to models trained directly on the input image.
Fichier principal
Vignette du fichier
ECCV2018.pdf (1.12 Mo) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01850921 , version 1 (28-07-2018)

Identifiers

  • HAL Id : hal-01850921 , version 1

Cite

Edouard Oyallon, Eugene Belilovsky, Sergey Zagoruyko, Michal Valko. Compressing the Input for CNNs with the First-Order Scattering Transform. ECCV 2018 - European Conference on Computer Vision, Sep 2018, Munich, Germany. ⟨hal-01850921⟩
386 View
446 Download

Share

Gmail Facebook X LinkedIn More