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

Edouard Oyallon 1, 2, 3 Eugene Belilovsky 4 Sergey Zagoruyko 5 Michal Valko 2
2 SEQUEL - Sequential Learning
Inria Lille - Nord Europe, CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189
5 WILLOW - Models of visual object recognition and scene understanding
Inria de Paris, DI-ENS - Département d'informatique de l'École normale supérieure
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.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [2 references]  Display  Hide  Download

https://hal.inria.fr/hal-01850921
Contributor : Michal Valko <>
Submitted on : Saturday, July 28, 2018 - 12:28:36 AM
Last modification on : Wednesday, April 17, 2019 - 12:15:36 PM
Document(s) archivé(s) le : Monday, October 29, 2018 - 12:41:06 PM

File

ECCV2018.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01850921, version 1

Citation

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

Share

Metrics

Record views

439

Files downloads

325