Pinterest Canvas: Large-Scale Image Generation at Pinterest DOI: XXXXXXX.XXXXXXX Conference: The 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining; August 9–13, 2026; Jeju, Korea ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computing methodologies Computer vision CCS: Computing methodologies Image representations CCS: Computing methodologies Neural networks
Yu Wang, Eric Tzeng, Raymond Shiau, Jie Yang, Dmitry Kislyuk, Charles Rosenberg · 2026
Read Pinterest Canvas: Large-Scale Image Generation at Pinterest DOI: XXXXXXX.XXXXXXX Conference: The 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining; August 9–13, 2026; Jeju, Korea ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computing methodologies Computer vision CCS: Computing methodologies Image representations CCS: Computing methodologies Neural networks by Yu Wang, Eric Tzeng, Raymond Shiau, Jie Yang, Dmitry Kislyuk, Charles Rosenberg (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-06453.txt.
Description
Wang and his team present Pinterest Canvas, an image generation system for editing and enhancement under strict rules. Canvas first trains a diffusion model on a multimodal dataset, then fine-tunes variants on task-specific datasets into expert models. The paper describes picking and inference ways, with background work and aspect-ratio fill as cases. Tests show double-digit engagement lifts, with 18.0 percent for background work; human raters prefer them to third-party models. Variants cover scene building from many images and making from image to video.
Catalog record
| CREATOR | Yu Wang, Eric Tzeng, Raymond Shiau, Jie Yang, Dmitry Kislyuk, Charles Rosenberg |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49516 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.06453