File Download
There are no files associated with this item.
Supplementary
-
Citations:
- Appears in Collections:
Conference Paper: Restoreformer: High-quality blind face restoration from undegraded key-value pairs
Title | Restoreformer: High-quality blind face restoration from undegraded key-value pairs |
---|---|
Authors | |
Issue Date | 2022 |
Publisher | IEEE Computer Society. |
Citation | IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (Virtual), New Orleans, Louisiana, USA, 19-24 June, 2022. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, p. 17512-17521 How to Cite? |
Abstract | Blind face restoration is to recover a high-quality face image from unknown degradations. As face image contains abundant contextual information, we propose a method, RestoreFormer, which explores fully-spatial attentions to model contextual information and surpasses existing works that use local convolutions. RestoreFormer has several benefits compared to prior arts. First, unlike the conventional multi-head self-attention in previous Vision Transformers (ViTs), RestoreFormer incorporates a multi-head cross-attention layer to learn fully-spatial interactions between corrupted queries and high-quality key-value pairs. Second, the key-value pairs in ResotreFormer are sampled from a reconstruction-oriented high-quality dictionary, whose elements are rich in high-quality facial features specifically aimed for face reconstruction, leading to superior restoration results. Third, RestoreFormer outperforms advanced state-of-the-art methods on one synthetic dataset and three real-world datasets, as well as produces images with better visual quality. |
Persistent Identifier | http://hdl.handle.net/10722/315796 |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Wang, Z | - |
dc.contributor.author | Zhang, J | - |
dc.contributor.author | Chen, R | - |
dc.contributor.author | Wang, W | - |
dc.contributor.author | Luo, P | - |
dc.date.accessioned | 2022-08-19T09:04:36Z | - |
dc.date.available | 2022-08-19T09:04:36Z | - |
dc.date.issued | 2022 | - |
dc.identifier.citation | IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (Virtual), New Orleans, Louisiana, USA, 19-24 June, 2022. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, p. 17512-17521 | - |
dc.identifier.uri | http://hdl.handle.net/10722/315796 | - |
dc.description.abstract | Blind face restoration is to recover a high-quality face image from unknown degradations. As face image contains abundant contextual information, we propose a method, RestoreFormer, which explores fully-spatial attentions to model contextual information and surpasses existing works that use local convolutions. RestoreFormer has several benefits compared to prior arts. First, unlike the conventional multi-head self-attention in previous Vision Transformers (ViTs), RestoreFormer incorporates a multi-head cross-attention layer to learn fully-spatial interactions between corrupted queries and high-quality key-value pairs. Second, the key-value pairs in ResotreFormer are sampled from a reconstruction-oriented high-quality dictionary, whose elements are rich in high-quality facial features specifically aimed for face reconstruction, leading to superior restoration results. Third, RestoreFormer outperforms advanced state-of-the-art methods on one synthetic dataset and three real-world datasets, as well as produces images with better visual quality. | - |
dc.language | eng | - |
dc.publisher | IEEE Computer Society. | - |
dc.relation.ispartof | Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022 | - |
dc.rights | Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Copyright © IEEE Computer Society. | - |
dc.title | Restoreformer: High-quality blind face restoration from undegraded key-value pairs | - |
dc.type | Conference_Paper | - |
dc.identifier.email | Luo, P: pluo@hku.hk | - |
dc.identifier.authority | Luo, P=rp02575 | - |
dc.identifier.hkuros | 335580 | - |
dc.identifier.spage | 17512 | - |
dc.identifier.epage | 17521 | - |
dc.publisher.place | United States | - |