InGAN: Capturing and Retargeting the “DNA” of a Natural Image
Generative Adversarial Networks (GANs) typically learn a distribution of images in a large image dataset, and are then able to generate new images from this distribution. However, each natural image has its own internal statistics, captured by its unique distribution of patches. In this paper we pro...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Institute of Electrical and Electronics Engineers (IEEE)
2021
|
| Online Access: | https://hdl.handle.net/1721.1/128957 |
| _version_ | 1826207028281868288 |
|---|---|
| author | Shocher, Assaf Bagon, Shai Isola, Phillip John Irani, Michal |
| author2 | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory |
| author_facet | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Shocher, Assaf Bagon, Shai Isola, Phillip John Irani, Michal |
| author_sort | Shocher, Assaf |
| collection | MIT |
| description | Generative Adversarial Networks (GANs) typically learn a distribution of images in a large image dataset, and are then able to generate new images from this distribution. However, each natural image has its own internal statistics, captured by its unique distribution of patches. In this paper we propose an ''Internal GAN'' (InGAN) - an image-specific GAN - which trains on a single input image and learns its internal distribution of patches. It is then able to synthesize a plethora of new natural images of significantly different sizes, shapes and aspect-ratios - all with the same internal patch-distribution (same ''DNA'') as the input image. In particular, despite large changes in global size/shape of the image, all elements inside the image maintain their local size/shape. InGAN is fully unsupervised, requiring no additional data other than the input image itself. Once trained on the input image, it can remap the input to any size or shape in a single feedforward pass, while preserving the same internal patch distribution. InGAN provides a unified framework for a variety of tasks, bridging the gap between textures and natural images. |
| first_indexed | 2024-09-23T13:42:21Z |
| format | Article |
| id | mit-1721.1/128957 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T13:42:21Z |
| publishDate | 2021 |
| publisher | Institute of Electrical and Electronics Engineers (IEEE) |
| record_format | dspace |
| spelling | mit-1721.1/1289572022-09-28T15:41:43Z InGAN: Capturing and Retargeting the “DNA” of a Natural Image Shocher, Assaf Bagon, Shai Isola, Phillip John Irani, Michal Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Generative Adversarial Networks (GANs) typically learn a distribution of images in a large image dataset, and are then able to generate new images from this distribution. However, each natural image has its own internal statistics, captured by its unique distribution of patches. In this paper we propose an ''Internal GAN'' (InGAN) - an image-specific GAN - which trains on a single input image and learns its internal distribution of patches. It is then able to synthesize a plethora of new natural images of significantly different sizes, shapes and aspect-ratios - all with the same internal patch-distribution (same ''DNA'') as the input image. In particular, despite large changes in global size/shape of the image, all elements inside the image maintain their local size/shape. InGAN is fully unsupervised, requiring no additional data other than the input image itself. Once trained on the input image, it can remap the input to any size or shape in a single feedforward pass, while preserving the same internal patch distribution. InGAN provides a unified framework for a variety of tasks, bridging the gap between textures and natural images. 2021-01-05T20:02:11Z 2021-01-05T20:02:11Z 2020-02 2019-10 2020-12-18T18:22:01Z Article http://purl.org/eprint/type/ConferencePaper 9781728148038 https://hdl.handle.net/1721.1/128957 Shocher, Assaf et al. "InGAN: Capturing and Retargeting the “DNA” of a Natural Image." Proceedings of the IEEE International Conference on Computer Vision, October-November 2019, Seoul, South Korea, Institute of Electrical and Electronics Engineers, February 2020. © 2019 IEEE en http://dx.doi.org/10.1109/iccv.2019.00459 Proceedings of the IEEE International Conference on Computer Vision Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf Institute of Electrical and Electronics Engineers (IEEE) arXiv |
| spellingShingle | Shocher, Assaf Bagon, Shai Isola, Phillip John Irani, Michal InGAN: Capturing and Retargeting the “DNA” of a Natural Image |
| title | InGAN: Capturing and Retargeting the “DNA” of a Natural Image |
| title_full | InGAN: Capturing and Retargeting the “DNA” of a Natural Image |
| title_fullStr | InGAN: Capturing and Retargeting the “DNA” of a Natural Image |
| title_full_unstemmed | InGAN: Capturing and Retargeting the “DNA” of a Natural Image |
| title_short | InGAN: Capturing and Retargeting the “DNA” of a Natural Image |
| title_sort | ingan capturing and retargeting the dna of a natural image |
| url | https://hdl.handle.net/1721.1/128957 |
| work_keys_str_mv | AT shocherassaf ingancapturingandretargetingthednaofanaturalimage AT bagonshai ingancapturingandretargetingthednaofanaturalimage AT isolaphillipjohn ingancapturingandretargetingthednaofanaturalimage AT iranimichal ingancapturingandretargetingthednaofanaturalimage |