Summary of Multi-style Facial Sketch Synthesis Through Masked Generative Modeling, by Bowen Sun et al.
Multi-Style Facial Sketch Synthesis through Masked Generative Modeling
by Bowen Sun, Guo Lu, Shibao Zheng
First submitted to arxiv on: 22 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed facial sketch synthesis (FSS) model can generate high-quality sketch portraits from given facial photographs, with applications in cross-modal face recognition, entertainment, art, and media. The model addresses challenges such as limited artist-drawn data, style constraints, and processing limitations by employing a lightweight end-to-end synthesis approach that converts images to multi-stylized sketches without requiring supplementary inputs. The study incorporates semi-supervised learning to overcome data insufficiency and uses feature extraction modules and style embeddings to guide the generative transformer during iterative prediction. The proposed method consistently outperforms previous algorithms across multiple benchmarks, demonstrating a significant disparity. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The FSS model can create sketch portraits from photos, which is important because it helps with face recognition, entertainment, art, and media. Right now, making high-quality sketches is hard because there’s not enough artist-drawn data, styles are limited, and processing information in existing models has flaws. To fix this, the study proposes a new model that converts images to multi-stylized sketches without needing extra help. The model uses semi-supervised learning to deal with lack of data and helps the generative transformer predict what to draw by extracting features and using style ideas. This method works better than previous ones on many tests. |
Keywords
» Artificial intelligence » Face recognition » Feature extraction » Semi supervised » Transformer
