StackGAN: Facial Image Generation Optimizations

TL;DR

This paper introduces a multi-stage StackGAN variant for facial image generation that aims to improve computational efficiency and training stability, generating grayscale images from edges and noise, with potential for broader application.

Contribution

The paper proposes a new multi-stage StackGAN architecture with conditional generators for facial image synthesis, focusing on reducing complexity and improving generalization.

Findings

Achieved FID scores of 73 for edge images and 59 for grayscale images.

Model shows potential for generalization despite subpar state-of-the-art results.

Dropout layers may reduce overfitting and improve performance.

Abstract

Current state-of-the-art photorealistic generators are computationally expensive, involve unstable training processes, and have real and synthetic distributions that are dissimilar in higher-dimensional spaces. To solve these issues, we propose a variant of the StackGAN architecture. The new architecture incorporates conditional generators to construct an image in many stages. In our model, we generate grayscale facial images in two different stages: noise to edges (stage one) and edges to grayscale (stage two). Our model is trained with the CelebA facial image dataset and achieved a Fr\'echet Inception Distance (FID) score of 73 for edge images and a score of 59 for grayscale images generated using the synthetic edge images. Although our model achieved subpar results in relation to state-of-the-art models, dropout layers could reduce the overfitting in our conditional mapping. Additionally, since most images can be broken down into important features, improvements to our model can generalize to other datasets. Therefore, our model can potentially serve as a superior alternative to traditional means of generating photorealistic images.