An instance mask representation for bubble size distribution in two-phase bubble flotation column based on deep learning model

IF 2.3 3区工程技术 Q2 ENGINEERING, MECHANICAL

Flow Measurement and Instrumentation Pub Date : 2025-03-19 DOI:10.1016/j.flowmeasinst.2025.102892

Zhiping Wen , Maiqiang Zhou , Sanja Mišković , Changchun Zhou

{"title":"An instance mask representation for bubble size distribution in two-phase bubble flotation column based on deep learning model","authors":"Zhiping Wen , Maiqiang Zhou , Sanja Mišković , Changchun Zhou","doi":"10.1016/j.flowmeasinst.2025.102892","DOIUrl":null,"url":null,"abstract":"<div><div>Measuring bubble size distribution from images sourced from various two-phase bubble systems presents a significant challenge, yet it holds substantial interest for many researchers in the field. This study introduces a novel approach by leveraging instance segmentation techniques based on deep learning to automatically quantify the bubble size distribution. The effective Mask RCNN and SOLO v2 were used as the basic model structure. The findings reveal that models employing the ResNet-FPN backbone outperform those using ResNet-C4/DC5 backbones in bubble segmentation. Specifically, the Mask RCNN with ResNet 101-FPN backbone achieved an average precision (AP) of 71.34 % for IoU = 0.50 and 68.13 % for IoU = 0.75. In terms of inference time, the SOLO v2 model displayed superior efficiency, taking 0.57 s per image compared to the Mask RCNN model. The study successfully demonstrated the utilization of the least squares fitting method to effectively detect and calculate bubble size distribution.</div></div>","PeriodicalId":50440,"journal":{"name":"Flow Measurement and Instrumentation","volume":"104 ","pages":"Article 102892"},"PeriodicalIF":2.3000,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Flow Measurement and Instrumentation","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0955598625000846","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

Measuring bubble size distribution from images sourced from various two-phase bubble systems presents a significant challenge, yet it holds substantial interest for many researchers in the field. This study introduces a novel approach by leveraging instance segmentation techniques based on deep learning to automatically quantify the bubble size distribution. The effective Mask RCNN and SOLO v2 were used as the basic model structure. The findings reveal that models employing the ResNet-FPN backbone outperform those using ResNet-C4/DC5 backbones in bubble segmentation. Specifically, the Mask RCNN with ResNet 101-FPN backbone achieved an average precision (AP) of 71.34 % for IoU = 0.50 and 68.13 % for IoU = 0.75. In terms of inference time, the SOLO v2 model displayed superior efficiency, taking 0.57 s per image compared to the Mask RCNN model. The study successfully demonstrated the utilization of the least squares fitting method to effectively detect and calculate bubble size distribution.

查看原文本刊更多论文

基于深度学习模型的两相气泡浮选柱气泡粒径分布实例掩模表示

从来自各种两相气泡系统的图像中测量气泡大小分布是一个重大的挑战，但它引起了该领域许多研究人员的极大兴趣。本研究引入了一种利用基于深度学习的实例分割技术来自动量化气泡大小分布的新方法。采用有效掩模RCNN和SOLO v2作为基本模型结构。结果表明，采用ResNet-FPN骨干网的模型在气泡分割方面优于使用ResNet-C4/DC5骨干网的模型。具体来说，具有ResNet 101-FPN主干的Mask RCNN在IoU = 0.50时的平均精度（AP）为71.34%，在IoU = 0.75时的平均精度（AP）为68.13%。在推理时间方面，与Mask RCNN模型相比，SOLO v2模型显示出更高的效率，每张图像仅需0.57 s。研究成功地证明了利用最小二乘拟合方法可以有效地检测和计算气泡尺寸分布。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Flow Measurement and Instrumentation 工程技术-工程：机械

CiteScore

4.30

自引率

13.60%

发文量

123

审稿时长

6 months

期刊介绍： Flow Measurement and Instrumentation is dedicated to disseminating the latest research results on all aspects of flow measurement, in both closed conduits and open channels. The design of flow measurement systems involves a wide variety of multidisciplinary activities including modelling the flow sensor, the fluid flow and the sensor/fluid interactions through the use of computation techniques; the development of advanced transducer systems and their associated signal processing and the laboratory and field assessment of the overall system under ideal and disturbed conditions. FMI is the essential forum for critical information exchange, and contributions are particularly encouraged in the following areas of interest: Modelling: the application of mathematical and computational modelling to the interaction of fluid dynamics with flowmeters, including flowmeter behaviour, improved flowmeter design and installation problems. Application of CAD/CAE techniques to flowmeter modelling are eligible. Design and development: the detailed design of the flowmeter head and/or signal processing aspects of novel flowmeters. Emphasis is given to papers identifying new sensor configurations, multisensor flow measurement systems, non-intrusive flow metering techniques and the application of microelectronic techniques in smart or intelligent systems. Calibration techniques: including descriptions of new or existing calibration facilities and techniques, calibration data from different flowmeter types, and calibration intercomparison data from different laboratories. Installation effect data: dealing with the effects of non-ideal flow conditions on flowmeters. Papers combining a theoretical understanding of flowmeter behaviour with experimental work are particularly welcome.