Learning to Hallucinate Face Images via Component Generation and Enhancement

TL;DR

This paper introduces a two-stage face hallucination approach that first generates facial components with CNNs and then enhances them by transferring detailed structures from high-resolution images, improving the quality of generated faces.

Contribution

The method uniquely combines component generation and detail enhancement to produce more accurate and detailed face images compared to existing techniques.

Findings

Outperforms state-of-the-art face hallucination methods

Effectively recovers fine facial details

Generates more realistic facial images

Abstract

We propose a two-stage method for face hallucination. First, we generate facial components of the input image using CNNs. These components represent the basic facial structures. Second, we synthesize fine-grained facial structures from high resolution training images. The details of these structures are transferred into facial components for enhancement. Therefore, we generate facial components to approximate ground truth global appearance in the first stage and enhance them through recovering details in the second stage. The experiments demonstrate that our method performs favorably against state-of-the-art methods