This Flask-based application generates sketches from textual descriptions using deep learning and generative models. It also transforms sketches into realistic facial images using advanced image translation techniques.
- Utilizes Multi-Head Self-Attention for text encoding.
- Employs a Conditional Variational Autoencoder (CVAE) conditioned on extracted attributes for generating sketches.
- Enhances sketches using a Conditional Generative Adversarial Network (CGAN) to produce high-fidelity facial details.
- Explores CycleGAN, Pix2Pix, and supervised models for sketch-to-face transformation.
- Achieves the best results with CycleGAN, generating highly realistic facial images from sketches.

