Ali R. Khojasteh, Willem van de Water, Jerry Westerweel
{"title":"Practical object and flow structure segmentation using artificial intelligence","authors":"Ali R. Khojasteh, Willem van de Water, Jerry Westerweel","doi":"10.1007/s00348-024-03852-7","DOIUrl":null,"url":null,"abstract":"<div><p>This paper explores integrating artificial intelligence (AI) segmentation models, particularly the Segment Anything Model (SAM), into fluid mechanics experiments. SAM’s architecture, comprising an image encoder, prompt encoder, and mask decoder, is investigated for its application in detecting and segmenting objects and flow structures. Additionally, we explore the integration of natural language prompts, such as BERT, to enhance SAM’s performance in segmenting specific objects. Through case studies, we found that SAM is robust in object detection in fluid experiments. However, segmentations related to flow properties, such as scalar turbulence and bubbly flows, require fine-tuning. To facilitate the application, we have established a repository (https://github.com/AliRKhojasteh/Flow_segmentation) where models and usage examples can be accessed.</p></div>","PeriodicalId":554,"journal":{"name":"Experiments in Fluids","volume":null,"pages":null},"PeriodicalIF":2.3000,"publicationDate":"2024-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s00348-024-03852-7.pdf","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Experiments in Fluids","FirstCategoryId":"5","ListUrlMain":"https://link.springer.com/article/10.1007/s00348-024-03852-7","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, MECHANICAL","Score":null,"Total":0}
引用次数: 0
Abstract
This paper explores integrating artificial intelligence (AI) segmentation models, particularly the Segment Anything Model (SAM), into fluid mechanics experiments. SAM’s architecture, comprising an image encoder, prompt encoder, and mask decoder, is investigated for its application in detecting and segmenting objects and flow structures. Additionally, we explore the integration of natural language prompts, such as BERT, to enhance SAM’s performance in segmenting specific objects. Through case studies, we found that SAM is robust in object detection in fluid experiments. However, segmentations related to flow properties, such as scalar turbulence and bubbly flows, require fine-tuning. To facilitate the application, we have established a repository (https://github.com/AliRKhojasteh/Flow_segmentation) where models and usage examples can be accessed.
本文探讨了如何将人工智能(AI)分割模型,特别是 "任意分割模型"(SAM),整合到流体力学实验中。SAM 的结构包括图像编码器、提示编码器和掩码解码器,本文研究了 SAM 在检测和分割物体及流动结构中的应用。此外,我们还探索了自然语言提示(如 BERT)的整合,以提高 SAM 在分割特定物体方面的性能。通过案例研究,我们发现 SAM 在流体实验中的物体检测方面非常稳健。然而,与标量湍流和气泡流等流动特性相关的分割需要进行微调。为了方便应用,我们建立了一个资源库(https://github.com/AliRKhojasteh/Flow_segmentation),在这里可以访问模型和使用示例。
期刊介绍:
Experiments in Fluids examines the advancement, extension, and improvement of new techniques of flow measurement. The journal also publishes contributions that employ existing experimental techniques to gain an understanding of the underlying flow physics in the areas of turbulence, aerodynamics, hydrodynamics, convective heat transfer, combustion, turbomachinery, multi-phase flows, and chemical, biological and geological flows. In addition, readers will find papers that report on investigations combining experimental and analytical/numerical approaches.