Collaging Class-specific GANs for Semantic Image Synthesis
By Yuheng Li, Yijun Li, Jingwan Lu, Eli Shechtman, Yong Jae Lee, Krishna Kumar Singh. 2021
36 citations recorded by the source corpus.
We propose a new approach for high resolution semantic image synthesis. It consists of one base image generator and multiple class-specific generators. The base generator generates high quality images based on a segmentation map. To further improve the quality of different objects, we create a bank of Generative Adversarial Networks (GANs) by separately training class-specific models. This has several benefits including – dedicated weights for each class; centrally aligned data for each model; additional training data from other sources, potential of higher resolution and quality; and easy manipulation of a specific object in the scene. Experiments show that our approach can generate high quality images in high resolution while having flexibility of object-level control by using class-specific generators.
Subfields: Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition, Generative Models
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Graphics and Computer-Aided Design
- Richard Zhang
- Eli Shechtman
- Alexei A. Efros
- Taesung Park
- Jingwan Lu
- Yong Jae Lee
- Yuheng Li
- Jianwei Yang
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Yong Jae Lee
- Ming-Hsuan Yang
- Dacheng Tao
- Trevor Darrell
- Xiaogang Wang
- Jitendra Malik
- Andrew Zisserman
In Generative Models
- Richard Zhang
- Eli Shechtman
- Alexei A. Efros
- Yong Jae Lee
- Taesung Park
- Jun-Yan Zhu
- Jingwan Lu
- Yuheng Li
Citations in the corpus
- GLIGEN: Open-Set Grounded Text-to-Image Generation (2023, background): “, text [65, 70, 80], box [61, 62, 78], semantic masks [36, 47], images [8,38,83].”
- Contrastive Learning for Diverse Disentangled Foreground Generation (2022, methodology): “Inspired by [26] to preserve better spatial alignment information with the input inpainting mask, once the feature reaches to 4 × 4 we upsample it to 16 × 16 with lateral connections.”
- BlobGAN: Spatially Disentangled Scene Representations (2022, background): “semantic segmentation maps [57, 44], object-attribute graphs [32, 8], text [99, 66, 52, 64, 69, 65], pose [42, 73, 3], and keypoints [11].”