Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generation and Editing of Mandrill Faces: Application to Sex Editing and Assessment

Nicolas M. Dibot, Julien P. Renoult, William Puech

TL;DR

The main novelty of proposed method is the ability to edit the sex of male or female mandrills by identifying a sex axis in the latent space of a specific GAN, and an assessment of the sex levels based on statistical features extracted from real image distributions.

Abstract

Generative AI has seen major developments in recent years, enhancing the realism of synthetic images, also known as computer-generated images. In addition, generative AI has also made it possible to modify specific image characteristics through image editing. Previous work has developed methods based on generative adversarial networks (GAN) for generating realistic images, in particular faces, but also to modify specific features. However, this work has never been applied to specific animal species. Moreover, the assessment of the results has been generally done subjectively, rather than quantitatively. In this paper, we propose an approach based on methods for generating images of faces of male or female mandrills, a non-human primate. The main novelty of proposed method is the ability to edit their sex by identifying a sex axis in the latent space of a specific GAN. In addition, we have developed an assessment of the sex levels based on statistical features extracted from real image distributions. The experimental results we obtained from a specific database are not only realistic, but also accurate, meeting a need for future work in behavioral experiments with wild mandrills.

Generation and Editing of Mandrill Faces: Application to Sex Editing and Assessment

TL;DR

The main novelty of proposed method is the ability to edit the sex of male or female mandrills by identifying a sex axis in the latent space of a specific GAN, and an assessment of the sex levels based on statistical features extracted from real image distributions.

Abstract

Generative AI has seen major developments in recent years, enhancing the realism of synthetic images, also known as computer-generated images. In addition, generative AI has also made it possible to modify specific image characteristics through image editing. Previous work has developed methods based on generative adversarial networks (GAN) for generating realistic images, in particular faces, but also to modify specific features. However, this work has never been applied to specific animal species. Moreover, the assessment of the results has been generally done subjectively, rather than quantitatively. In this paper, we propose an approach based on methods for generating images of faces of male or female mandrills, a non-human primate. The main novelty of proposed method is the ability to edit their sex by identifying a sex axis in the latent space of a specific GAN. In addition, we have developed an assessment of the sex levels based on statistical features extracted from real image distributions. The experimental results we obtained from a specific database are not only realistic, but also accurate, meeting a need for future work in behavioral experiments with wild mandrills.
Paper Structure (19 sections, 20 figures, 2 algorithms)

This paper contains 19 sections, 20 figures, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related work
  3. Generative AI, GAN and encoding
  4. Handmade versus GAN-based editing
  5. Deep learning applied to mandrill behavior and visual communication
  6. The proposed method for editing images of mandrill faces
  7. Overview of the training phase
  8. Generation of mandrill faces from StyleGAN3-mandrill
  9. Training of the pSp-mandrill encoder
  10. The editing vector calculation
  11. Overview of the editing phase
  12. Possibility of the sex level deviation
  13. Calculation and assessment of the sex level from the space $W$ to to the space $W+$ of an edited generated image of a mandrill face
  14. Experimental results
  15. The Mandrillus Face Database TIEO2023108939
  16. ...and 4 more sections

Figures (20)

  • Figure 1: Overview of the training phase.
  • Figure 2: Calculation of the editing vector from SVM on the clusters of female and male vectors in the space $W$.
  • Figure 3: Overview of the editing phase.
  • Figure 4: Calculation of the upper and lower edit bounds for each sex according to the distributions of real images projected on the sex axis.
  • Figure 5: Example of the research of the optimal editing step $\Delta_e$ for editing the sex level of a mandrill face. Starting from the sex level $S_0$ of the original generated image, to reach the sex level $S_R$, in the first step, $\Delta_1$ corresponds to condition $C$ of the Alg. \ref{['algo:alg2']}, for the second step $\Delta_2$ to condition $B$, while for the third step $\Delta_3$ to condition $A$ and finally $\Delta_4$ to the optimal $\Delta_e$.
  • ...and 15 more figures