Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2312.06101 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929337413926912 |
|---|---|
| author | Huang, Binxiao Li, Jason Chun Lok Ran, Jie Li, Boyu Zhou, Jiajun Yu, Dahai Wong, Ngai |
| author_facet | Huang, Binxiao Li, Jason Chun Lok Ran, Jie Li, Boyu Zhou, Jiajun Yu, Dahai Wong, Ngai |
| contents | Conventional super-resolution (SR) schemes make heavy use of convolutional neural networks (CNNs), which involve intensive multiply-accumulate (MAC) operations, and require specialized hardware such as graphics processing units. This contradicts the regime of edge AI that often runs on devices strained by power, computing, and storage resources. Such a challenge has motivated a series of lookup table (LUT)-based SR schemes that employ simple LUT readout and largely elude CNN computation. Nonetheless, the multi-megabyte LUTs in existing methods still prohibit on-chip storage and necessitate off-chip memory transport. This work tackles this storage hurdle and innovates hundred-kilobyte LUT (HKLUT) models amenable to on-chip cache. Utilizing an asymmetric two-branch multistage network coupled with a suite of specialized kernel patterns, HKLUT demonstrates an uncompromising performance and superior hardware efficiency over existing LUT schemes. Our implementation is publicly available at: https://github.com/jasonli0707/hklut. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2312_06101 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Hundred-Kilobyte Lookup Tables for Efficient Single-Image Super-Resolution Huang, Binxiao Li, Jason Chun Lok Ran, Jie Li, Boyu Zhou, Jiajun Yu, Dahai Wong, Ngai Image and Video Processing Computer Vision and Pattern Recognition Conventional super-resolution (SR) schemes make heavy use of convolutional neural networks (CNNs), which involve intensive multiply-accumulate (MAC) operations, and require specialized hardware such as graphics processing units. This contradicts the regime of edge AI that often runs on devices strained by power, computing, and storage resources. Such a challenge has motivated a series of lookup table (LUT)-based SR schemes that employ simple LUT readout and largely elude CNN computation. Nonetheless, the multi-megabyte LUTs in existing methods still prohibit on-chip storage and necessitate off-chip memory transport. This work tackles this storage hurdle and innovates hundred-kilobyte LUT (HKLUT) models amenable to on-chip cache. Utilizing an asymmetric two-branch multistage network coupled with a suite of specialized kernel patterns, HKLUT demonstrates an uncompromising performance and superior hardware efficiency over existing LUT schemes. Our implementation is publicly available at: https://github.com/jasonli0707/hklut. |
| title | Hundred-Kilobyte Lookup Tables for Efficient Single-Image Super-Resolution |
| topic | Image and Video Processing Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2312.06101 |