UNet--: Memory-Efficient and Feature-Enhanced Network Architecture based on U-Net with Reduced Skip-Connections
保存先:
| 出版年: | arXiv.org (Dec 24, 2024), p. n/a |
|---|---|
| 第一著者: | |
| その他の著者: | , , , , , |
| 出版事項: |
Cornell University Library, arXiv.org
|
| 主題: | |
| オンライン･アクセス: | Citation/Abstract Full text outside of ProQuest |
| タグ: |
タグなし, このレコードへの初めてのタグを付けませんか!
|
MARC
| LEADER | 00000nab a2200000uu 4500 | ||
|---|---|---|---|
| 001 | 3149108798 | ||
| 003 | UK-CbPIL | ||
| 022 | |a 2331-8422 | ||
| 024 | 7 | |a 10.1007/978-981-96-0963-5 |2 doi | |
| 035 | |a 3149108798 | ||
| 045 | 0 | |b d20241224 | |
| 100 | 1 | |a Yin, Lingxiao | |
| 245 | 1 | |a UNet--: Memory-Efficient and Feature-Enhanced Network Architecture based on U-Net with Reduced Skip-Connections | |
| 260 | |b Cornell University Library, arXiv.org |c Dec 24, 2024 | ||
| 513 | |a Working Paper | ||
| 520 | 3 | |a U-Net models with encoder, decoder, and skip-connections components have demonstrated effectiveness in a variety of vision tasks. The skip-connections transmit fine-grained information from the encoder to the decoder. It is necessary to maintain the feature maps used by the skip-connections in memory before the decoding stage. Therefore, they are not friendly to devices with limited resource. In this paper, we propose a universal method and architecture to reduce the memory consumption and meanwhile generate enhanced feature maps to improve network performance. To this end, we design a simple but effective Multi-Scale Information Aggregation Module (MSIAM) in the encoder and an Information Enhancement Module (IEM) in the decoder. The MSIAM aggregates multi-scale feature maps into single-scale with less memory. After that, the aggregated feature maps can be expanded and enhanced to multi-scale feature maps by the IEM. By applying the proposed method on NAFNet, a SOTA model in the field of image restoration, we design a memory-efficient and feature-enhanced network architecture, UNet--. The memory demand by the skip-connections in the UNet-- is reduced by 93.3%, while the performance is improved compared to NAFNet. Furthermore, we show that our proposed method can be generalized to multiple visual tasks, with consistent improvements in both memory consumption and network accuracy compared to the existing efficient architectures. | |
| 653 | |a Visual tasks | ||
| 653 | |a Feature maps | ||
| 653 | |a Image restoration | ||
| 653 | |a Memory tasks | ||
| 653 | |a Memory devices | ||
| 653 | |a Modules | ||
| 653 | |a Decoding | ||
| 653 | |a Consumption | ||
| 653 | |a Coders | ||
| 653 | |a Effectiveness | ||
| 700 | 1 | |a Tao, Wei | |
| 700 | 1 | |a Zhao, Dongyue | |
| 700 | 1 | |a Ito, Tadayuki | |
| 700 | 1 | |a Osa, Kinya | |
| 700 | 1 | |a Kato, Masami | |
| 700 | 1 | |a Tse-Wei, Chen | |
| 773 | 0 | |t arXiv.org |g (Dec 24, 2024), p. n/a | |
| 786 | 0 | |d ProQuest |t Engineering Database | |
| 856 | 4 | 1 | |3 Citation/Abstract |u https://www.proquest.com/docview/3149108798/abstract/embedded/ZKJTFFSVAI7CB62C?source=fedsrch |
| 856 | 4 | 0 | |3 Full text outside of ProQuest |u http://arxiv.org/abs/2412.18276 |