Boundless: Generative Adversarial Networks for Image Extension

By Piotr Teterwak, Aaron Sarna, Dilip Krishnan, Aaron Maschinot, David Belanger, Ce Liu, William T. Freeman. 2019

131 citations recorded by the source corpus.

Image extension models have broad applications in image editing, computational photography and computer graphics. While image inpainting has been extensively studied in the literature, it is challenging to directly apply the state-of-the-art inpainting methods to image extension as they tend to generate blurry or repetitive pixels with inconsistent semantics. We introduce semantic conditioning to the discriminator of a generative adversarial network (GAN), and achieve strong results on image extension with coherent semantics and visually pleasing colors and textures. We also show promising results in extreme extensions, such as panorama generation.

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

1. Ce Liu
2. Michal Irani
3. William T. Freeman
4. Jonathan Ho
5. Ming-Hsuan Yang
6. Dilip Krishnan
7. Piotr Teterwak
8. Hung-Yu Tseng

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Ming-Hsuan Yang
3. Dacheng Tao
4. Andrea Vedaldi
5. Xiaogang Wang
6. Jitendra Malik
7. Trevor Darrell
8. Andrew Zisserman

In Generative Models

1. Ming-Hsuan Yang
2. Jonathan Ho
3. David J. Fleet
4. Mohammad Norouzi
5. Chitwan Saharia
6. Lu Jiang
7. William Chan
8. Tim Salimans

Citations in the corpus

• Palette: Image-to-Image Diffusion Models (2021, methodology): “, 2021] and Boundless [Teterwak et al., 2019] on scenery categories.”
• Regularizing Generative Adversarial Networks under Limited Data (2021, background): “The GAN models are the cornerstone techniques for numerous vision applications, such as data augmentation [11, 12], domain adaptation [18, 19], image extrapolation [60], image-to-image translation [20, 34, 75], and image editing [1, 4, 21, 63].”
• Understanding invariance via feedforward inversion of discriminatively trained classifiers (2021, methodology): “Such feature-based conditioning of the discriminator is similar to Boundless (Teterwak et al., 2019), and the feature conditioning of the generator is similar to SPADE (Park et al.”
• SynSin: End-to-End View Synthesis From a Single Image (2019, methodology): “Deep networks have been previously applied to inpainting [46, 61, 66].”
• Semantic Pyramid for Image Generation (2020, methodology): “, [4, 37, 33]), we utilize the information available in deep features for that region.”
• InfinityGAN: Towards Infinite-Resolution Image Synthesis (2021, methodology): “In Table 2, our model performs favorably against all baselines in image outpainting (Boundless [34], NS-outpaint [40], and In&Out [4]).”
• InOut: Diverse Image Outpainting via GAN Inversion (2021, methodology): “Similar to [27], we evaluate our method on a subset of the Places365 dataset.”