[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-290360-105":59,"doc-detail-290360-en":131},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","machine-learning-methods-for-droplet-analysis-bachelors-thesis","Machine Learning Methods for Droplet Analysis - Bachelor’s Thesis","","Measuring wetting-related surface properties is essential for industrial and scientific applications, particularly digital microfluidics, where electrode grids require surfaces with high hydrophobicity and low surface heterogeneity. The standard contact angle goniometry method lacks throughput and is unsuitable for on-line production testing. This thesis develops machine learning models for droplet-based wetting analysis. A custom CNN was trained on synthetic data to localise droplets, then compared with fine-tuned YOLOv5 bounding-box detection. YOLOv5 outputs across experimental frames were converted to masks to drive a ResNet model predicting surface heterogeneity directly, enabling high-throughput characterisation.",{"@graph":69,"@context":123},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/machine-learning-methods-for-droplet-analysis-bachelors-thesis/290360/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/machine-learning-methods-for-droplet-analysis-bachelors-thesis/290360.png","ImageObject",300,407,{"name":92,"@type":93},"Eliana","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-28","2026-09-17",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",12,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"Why is measuring wetting-related surface properties important in this thesis?","Question",{"text":113,"@type":114},"Wetting characteristics such as hydrophobicity and surface heterogeneity are critical for applications like digital microfluidics, where droplet manipulation depends on suitable surface properties.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"What limitation of contact angle goniometry motivates the proposed work?",{"text":118,"@type":114},"Contact angle goniometry has low throughput and is impractical for implementation at the end of a production line, motivating faster measurement approaches.",{"name":120,"@type":111,"acceptedAnswer":121},"Which machine learning models are developed and how are they used?",{"text":122,"@type":114},"The thesis develops a custom CNN for droplet localisation, compares it against fine-tuned YOLOv5 for bounding-box detection, and then uses YOLOv5 mask outputs from experimental frames to train a ResNet-based model to predict surface heterogeneity.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},290360,1789641480,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":36,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":130,"read_time":144,"preview_page_count":4},4398048949847,"https://ap-avatar.wpscdn.com/avatar/400002536579ef2da7f?_k=1778318612642679267","Bachelor’s Programme in Science and Technology  \nMachine Learning Methods for Droplet Analysis  \nComparison of Machine Learning Methods for Macroscopic Water Droplet Tracking and a Method for Generating Contact Angle Hysteresis Heatmaps  \nMiika Vuorio  \nBACHELOR’S THESIS  \nAalto University  \nBACHELOR’S THESIS 2025  \nMachine Learning Methods for Droplet Analysis  \nComparison of Machine Learning Methods for Macroscopic Water Droplet Tracking and a Method for Generating Contact Angle Hysteresis Heatmaps  \nMiika Vuorio  \nThesis submitted in partial fulfillment of the requirements for  \nthe degree of Bachelor of Science in Technology.  \nOtaniemi, 12 Dec, 2025  \nSupervisor: Professor Robin Ras  \nAdvisor: Research Fellow Juuso Korhonen  \nAalto University  \nSchool of Science  \nBachelor’s Programme in Science and Technology  \nAbstract  \nAalto University, P.O. Box 11000, FI-00076 Aalto [www.aalto.fi](www.aalto.fi)  \nAuthor  \nMiika Vuorio  \n\n| Title\u003Cbr>Machine learning methods for droplet analysis |\n| --- |\n| School School of Science |\n| Degree programme Bachelor’s Programme in Science and Technology |\n| Major Quantum Technology Code SCI3103 |\n| Supervisor Robin Ras |\n| Advisor Juuso Korhonen |\n| Level Bachelor’s thesis Date 12 Dec 2025 Pages 40 Language English |\n\nAbstract  \nMeasuring wetting-related surface properties is essential for many industrial and scientific applications. For example, in the active research field of digital microfluidics, where electrode grids are employed to efficiently and precisely control the location of a droplet, candidate surfaces must exhibit high hydrophobicity and low surface heterogeneity. The current standard methodology for characterising these properties—contact angle goniometry—suffers from low throughput and is infeasible to implement at the end of a production line, e.g., for the aforementioned digital microfluidic chips.  \nConsequently, developing more efficient methods for measuring surface properties of materials would facilitate industrial scalability. Despite the recent success of machine learning techniques in various scientific and industrial applications, their use in analysing wetting characteristics of materials has remained relatively unexplored. To address this need, this thesis develops machine learning models for analysing surface wetting.  \nFirst, a custom convolutional neural network (CNN) for localising droplets on a surface was developed and trained on synthetic data. The performance of this model was then compared to a contemporary object recognition model, YOLOv5, that was fine-tuned to draw bounding boxes around droplets. Finally, outputs of YOLOv5 from 24 frames of experimental video were converted into masksand combined with the original video to serve as input to a ResNet-based model trained to predict surface heterogeneity directly.  \nThe custom CNN achieved reasonable accuracy with a mean Euclidean distance of 5.34 pixels between the predicted droplet centre and the ground truth on the synthetic dataset. However, it was outperformed by the fine-tuned YOLOv5 model, which had a mean Euclidean error of 1.31 pixels and more robust performance on experimental data. Finally, the ResNet-based model was able to characterise large differences in surface heterogeneity but lacked the sensitivity to distinguish between materials with a small difference in contact angle hysteresis. These results demonstrate the feasibility of machine learning methods for high-throughput material characterisation.  \nKeywords Contact angle hysteresis, droplet tracking, machine learning, wetting  \nurn [https://aaltodoc.aalto.fi](https://aaltodoc.aalto.fi)  \nContents  \nAbstract ii  \nContents iii  \n1 . Introduction 1  \n2 . Theoretical Background 3  \n2.1 Theoretical Framework for Inferring Material Properties from Droplet Video ............ 3  \n2.1.1 Advancing and Receding Contact Angles and the Pinning Force ...................... 3  \n2.1.2 The Furmidge Equation ................ 4  \n2.1.3 Geometry of a Moving D","cbCaisvFw1plMFv1","https://ap.wps.com/l/cbCaisvFw1plMFv1","pdf",8524566,"English","# Abstract\n# Contents\n# 1. Introduction\n# 2. Theoretical Background\n## 2.1 Theoretical Framework for Inferring Material Properties from Droplet Video\n## 2.2 Neural Network Architectures\n## 2.3 The Architecture of YOLOv5\n# 3. Methods\n## 3.1 The Data\n## 3.2 Convolutional Neural Network\n## 3.3 Fine-tuning the YOLOv5 Head\n## 3.4 ResNet-Based Machine Learning Network for Creating CAH Heatmaps\n# 4. Results\n## 4.1 Custom Convolutional Model Results\n## 4.2 YOLOv5 Results\n## 4.3 CAH Heatmap Model Results\n# 5. Discussion\n## 5.1 Future Directions for Droplet Position Inference Models\n## 5.2 Future Directions for CAH Inference Models\n# Bibliography\n# A. Details on Code Repository","[{\"question\":\"Why is measuring wetting-related surface properties important in this thesis?\",\"answer\":\"Wetting characteristics such as hydrophobicity and surface heterogeneity are critical for applications like digital microfluidics, where droplet manipulation depends on suitable surface properties.\"},{\"question\":\"What limitation of contact angle goniometry motivates the proposed work?\",\"answer\":\"Contact angle goniometry has low throughput and is impractical for implementation at the end of a production line, motivating faster measurement approaches.\"},{\"question\":\"Which machine learning models are developed and how are they used?\",\"answer\":\"The thesis develops a custom CNN for droplet localisation, compares it against fine-tuned YOLOv5 for bounding-box detection, and then uses YOLOv5 mask outputs from experimental frames to train a ResNet-based model to predict surface heterogeneity.\"}]","Machine Learning Methods for Droplet Analysis - Bachelor’s Thesis | PDF",101]