Abstract

This paper provides information on how to colorize grayscale or sketch images using VAE. It also provides a brief overview of color spaces in the form of l*a*b* where l* for lightness, a* is the red/green coordinate, and b* is the yellow/blue coordinate. There are two types of grayscale colorization. Grayscale photos are transformed into colorful images through the process of colorization in a way that is visually appealing. The main goal is to convince the user that the result is legitimate. This paper suggests a flexible VAE-based image colorization technique. The VAE is applied to learn how images are represented in color. The VAE will train the model with both adversarial loss and perceptual loss on the discriminator, resulting in a more accurate learning of colors and more appropriate color outcomes. This study explores a rapid and efficient method for turning grayscale and sketch shots into color images with the aid of an existing color scheme