| Title: |
Data-dependent scaling of CNN's first layer for improved image manipulation detection |
| Authors: |
Castillo Camacho, Ivan; Wang, Kai |
| Contributors: |
GIPSA - Apprentissage, Classification, Traitement d'Images et de Vidéos (GIPSA-ACTIV); GIPSA Pôle Sciences des Données (GIPSA-PSD); Grenoble Images Parole Signal Automatique (GIPSA-lab); Centre National de la Recherche Scientifique (CNRS)-Université Grenoble Alpes (UGA)-Institut polytechnique de Grenoble - Grenoble Institute of Technology (Grenoble INP); Université Grenoble Alpes (UGA)-Centre National de la Recherche Scientifique (CNRS)-Université Grenoble Alpes (UGA)-Institut polytechnique de Grenoble - Grenoble Institute of Technology (Grenoble INP); Université Grenoble Alpes (UGA)-Grenoble Images Parole Signal Automatique (GIPSA-lab); Université Grenoble Alpes (UGA) |
| Source: |
IWDW 2020 - 19th International Workshop on Digital-forensics and Watermarking ; https://hal.science/hal-03000629 ; IWDW 2020 - 19th International Workshop on Digital-forensics and Watermarking, Nov 2020, Melbourne (online), Australia. pp.208-223, ⟨10.1007/978-3-030-69449-4_16⟩ ; http://iwdw.site/ |
| Publisher Information: |
CCSD; Springer |
| Publication Year: |
2020 |
| Collection: |
Université Grenoble Alpes: HAL |
| Subject Terms: |
[INFO.INFO-MM]Computer Science [cs]/Multimedia [cs.MM] |
| Subject Geographic: |
Melbourne (online); Australia |
| Description: |
International audience ; Convolutional Neural Networks (CNNs) have become an effective tool to detect image manipulation operations, e.g., noise addition, median filtering and JPEG compression. In this paper, we propose a simple and practical method for adjusting the CNN's first layer, based on a proper scaling of first-layer filters with a data-dependent approach. The key idea is to keep the stability of the variance of data flow in a CNN. We also present studies on the output variance for convolutional filter, which are the basis of our proposed scaling. The proposed method can cope well with different first-layer initialization algorithms and different CNN architectures. The experiments are performed with two challenging forensic problems, i.e., a multi-class classification problem of a group of manipulation operations and a binary detection problem of JPEG compression with high quality factor, both on relatively small image patches. Experimental results show the utility of our method with a noticeable and consistent performance improvement after scaling. |
| Document Type: |
conference object |
| Language: |
English |
| DOI: |
10.1007/978-3-030-69449-4_16 |
| Availability: |
https://hal.science/hal-03000629; https://hal.science/hal-03000629v1/document; https://hal.science/hal-03000629v1/file/IWDW20.pdf; https://doi.org/10.1007/978-3-030-69449-4_16 |
| Rights: |
info:eu-repo/semantics/OpenAccess |
| Accession Number: |
edsbas.84EE2EFB |
| Database: |
BASE |