Learning Based Image Coloring

The project "Learning Based Image Coloring" presents an innovative end-to-end system employing deep neural networks to effectively colorize grayscale images. Utilizing General Adversarial Networks (GANs), two models—generative and discriminative—are trained concurrently. The process involves inputting a grayscale image into the GAN, which then generates a colored version of the original. Through training, a Discriminative network discerns real from false images, while the GAN endeavors to deceive it by producing convincing colors. This dual-network approach aims to construct a robust and precise system for automated grayscale image colorization, promising significant advancements in image processing and enhancement.