Paying U-Attention to Textures: Multi-Stage Hourglass Vision Transformer for Universal Texture Synthesis
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2022
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914904447909888 |
|---|---|
| author | Guo, Shouchang Deschaintre, Valentin Noll, Douglas Roullier, Arthur |
| author_facet | Guo, Shouchang Deschaintre, Valentin Noll, Douglas Roullier, Arthur |
| contents | We present a novel U-Attention vision Transformer for universal texture synthesis. We exploit the natural long-range dependencies enabled by the attention mechanism to allow our approach to synthesize diverse textures while preserving their structures in a single inference. We propose a hierarchical hourglass backbone that attends to the global structure and performs patch mapping at varying scales in a coarse-to-fine-to-coarse stream. Completed by skip connection and convolution designs that propagate and fuse information at different scales, our hierarchical U-Attention architecture unifies attention to features from macro structures to micro details, and progressively refines synthesis results at successive stages. Our method achieves stronger 2$\times$ synthesis than previous work on both stochastic and structured textures while generalizing to unseen textures without fine-tuning. Ablation studies demonstrate the effectiveness of each component of our architecture. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2202_11703 |
| institution | arXiv |
| publishDate | 2022 |
| record_format | arxiv |
| spellingShingle | Paying U-Attention to Textures: Multi-Stage Hourglass Vision Transformer for Universal Texture Synthesis Guo, Shouchang Deschaintre, Valentin Noll, Douglas Roullier, Arthur Computer Vision and Pattern Recognition Graphics Image and Video Processing We present a novel U-Attention vision Transformer for universal texture synthesis. We exploit the natural long-range dependencies enabled by the attention mechanism to allow our approach to synthesize diverse textures while preserving their structures in a single inference. We propose a hierarchical hourglass backbone that attends to the global structure and performs patch mapping at varying scales in a coarse-to-fine-to-coarse stream. Completed by skip connection and convolution designs that propagate and fuse information at different scales, our hierarchical U-Attention architecture unifies attention to features from macro structures to micro details, and progressively refines synthesis results at successive stages. Our method achieves stronger 2$\times$ synthesis than previous work on both stochastic and structured textures while generalizing to unseen textures without fine-tuning. Ablation studies demonstrate the effectiveness of each component of our architecture. |
| title | Paying U-Attention to Textures: Multi-Stage Hourglass Vision Transformer for Universal Texture Synthesis |
| topic | Computer Vision and Pattern Recognition Graphics Image and Video Processing |
| url | https://arxiv.org/abs/2202.11703 |