用于生成复杂全息图的量化神经网络

Yutaka Endo, Minoru Oikawa, Timothy D. Wilkinson, Tomoyoshi Shimobaba, Tomoyoshi Ito
{"title":"用于生成复杂全息图的量化神经网络","authors":"Yutaka Endo, Minoru Oikawa, Timothy D. Wilkinson, Tomoyoshi Shimobaba, Tomoyoshi Ito","doi":"arxiv-2409.06711","DOIUrl":null,"url":null,"abstract":"Computer-generated holography (CGH) is a promising technology for augmented\nreality displays, such as head-mounted or head-up displays. However, its high\ncomputational demand makes it impractical for implementation. Recent efforts to\nintegrate neural networks into CGH have successfully accelerated computing\nspeed, demonstrating the potential to overcome the trade-off between\ncomputational cost and image quality. Nevertheless, deploying neural\nnetwork-based CGH algorithms on computationally limited embedded systems\nrequires more efficient models with lower computational cost, memory footprint,\nand power consumption. In this study, we developed a lightweight model for\ncomplex hologram generation by introducing neural network quantization.\nSpecifically, we built a model based on tensor holography and quantized it from\n32-bit floating-point precision (FP32) to 8-bit integer precision (INT8). Our\nperformance evaluation shows that the proposed INT8 model achieves hologram\nquality comparable to that of the FP32 model while reducing the model size by\napproximately 70% and increasing the speed fourfold. Additionally, we\nimplemented the INT8 model on a system-on-module to demonstrate its\ndeployability on embedded platforms and high power efficiency.","PeriodicalId":501174,"journal":{"name":"arXiv - CS - Graphics","volume":"13 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Quantized neural network for complex hologram generation\",\"authors\":\"Yutaka Endo, Minoru Oikawa, Timothy D. Wilkinson, Tomoyoshi Shimobaba, Tomoyoshi Ito\",\"doi\":\"arxiv-2409.06711\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Computer-generated holography (CGH) is a promising technology for augmented\\nreality displays, such as head-mounted or head-up displays. However, its high\\ncomputational demand makes it impractical for implementation. Recent efforts to\\nintegrate neural networks into CGH have successfully accelerated computing\\nspeed, demonstrating the potential to overcome the trade-off between\\ncomputational cost and image quality. Nevertheless, deploying neural\\nnetwork-based CGH algorithms on computationally limited embedded systems\\nrequires more efficient models with lower computational cost, memory footprint,\\nand power consumption. In this study, we developed a lightweight model for\\ncomplex hologram generation by introducing neural network quantization.\\nSpecifically, we built a model based on tensor holography and quantized it from\\n32-bit floating-point precision (FP32) to 8-bit integer precision (INT8). Our\\nperformance evaluation shows that the proposed INT8 model achieves hologram\\nquality comparable to that of the FP32 model while reducing the model size by\\napproximately 70% and increasing the speed fourfold. Additionally, we\\nimplemented the INT8 model on a system-on-module to demonstrate its\\ndeployability on embedded platforms and high power efficiency.\",\"PeriodicalId\":501174,\"journal\":{\"name\":\"arXiv - CS - Graphics\",\"volume\":\"13 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-08-25\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Graphics\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.06711\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Graphics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.06711","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

计算机生成全息(CGH)是一种很有前途的增强现实显示技术,例如头戴式或平视显示器。然而,它对计算能力的高要求使其无法付诸实施。最近,将神经网络集成到 CGH 中的努力成功地加快了计算速度,证明了克服计算成本与图像质量之间权衡的潜力。然而,在计算能力有限的嵌入式系统上部署基于神经网络的 CGH 算法需要更高效的模型、更低的计算成本、内存占用和功耗。具体来说,我们建立了一个基于张量全息的模型,并将其从 32 位浮点精度(FP32)量化为 8 位整数精度(INT8)。我们的性能评估结果表明,所提出的 INT8 模型实现了与 FP32 模型相当的全息质量,同时将模型大小减少了约 70%,速度提高了四倍。此外,我们还在一个系统模块上实现了 INT8 模型,以证明它在嵌入式平台上的可部署性和高能效。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Quantized neural network for complex hologram generation
Computer-generated holography (CGH) is a promising technology for augmented reality displays, such as head-mounted or head-up displays. However, its high computational demand makes it impractical for implementation. Recent efforts to integrate neural networks into CGH have successfully accelerated computing speed, demonstrating the potential to overcome the trade-off between computational cost and image quality. Nevertheless, deploying neural network-based CGH algorithms on computationally limited embedded systems requires more efficient models with lower computational cost, memory footprint, and power consumption. In this study, we developed a lightweight model for complex hologram generation by introducing neural network quantization. Specifically, we built a model based on tensor holography and quantized it from 32-bit floating-point precision (FP32) to 8-bit integer precision (INT8). Our performance evaluation shows that the proposed INT8 model achieves hologram quality comparable to that of the FP32 model while reducing the model size by approximately 70% and increasing the speed fourfold. Additionally, we implemented the INT8 model on a system-on-module to demonstrate its deployability on embedded platforms and high power efficiency.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信