Generating realistic faces from sketch images or textual descriptions is a fundamental challenge in computer vision due to the limited facial details in sketches. this has a face hallucination problem. It involves an image translation network utilizing adversarial networks to enhance facial attribute accuracy. It combines sketch images and attribute features perceptually, distinguishing it from most attribute-embedded networks. network comprises a feature extractor and down-sampling/up-sampling networks, using skip-connections to reduce layer complexity while maintaining performance. The discriminator assesses attribute presence in generated faces. This approach outperforms current image translation methods.

• Adversarial Networks
• Computer Vision
• Image Translation