[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-120111-en":3,"doc-seo-120111-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},120111,8796095462418,"Noah","https://ap-avatar.wpscdn.com/avatar/80000253c1241d02b47?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778826106357471780",8,"Research & Report","Improving and Evaluating Machine Learning Methods for Forensic Shoeprint Matching","A machine learning pipeline is proposed to enhance accuracy and generalisability in forensic shoeprint pattern matching. The method extracts 2D coordinates from shoeprint scans via edge detection, aligns paired prints using iterative closest point (ICP), and computes similarity metrics. A random forest is trained on these metrics to output a probabilistic measure of whether two prints likely originate from the same outsole. Generalisability is tested on partial prints, blur, wear, and different shoe models, showing poor cross-scenario transfer from single-scenario training.","arXiv :2405 . 14878v1 [ ee ss .IV] 2 Apr 2024  \nJournal Title Here, 2023, pp. 1–20  \ndoi: DOI HERE  \nAdvance Access Publication Date: Day Month Year Paper  \nPAPER  \nImproving and Evaluating Machine Learning Methods for Forensic Shoeprint Matching  \nDivij Jain , 1,† Saatvik Kher ,2,† Lena Liang ,3,† Yufeng Wu , 1,† Ashley Zheng ,4,† Xizhen Cai, 1 Anna Plantinga 1 and Elizabeth Upton1,∗  \n1 Department of Mathematics and Statistics, Williams College, 15 Hoxsey Street, Williamstown, MA USA, 01267, 2 Department of Mathematics and Statistics, Pomona College, CA, USA, 3 Department of Statistics, University of Chicago, Il, USA and 4 Department of Mathematics, Fairfield University, CT, USA  \n†Equal Contribution∗ Corresponding [author. emu1@williams.edu](author. emu1@williams.edu)  \nAbstract  \nWe propose a machine learning pipeline for forensic shoeprint pattern matching that improves on the accuracy and generalisability of existing methods. We extract 2D coordinates from shoeprint scans using edge detection and align the two shoeprints with iterative closest point (ICP) . We then extract similarity metrics to quantify how well the two prints match and use these metrics to train a random forest that generates a probabilistic measurement of how likely two prints are to have originated from the same outsole. We assess the generalisability of machine learning methods trained on lab shoeprint scans to more realistic crime scene shoeprint data by evaluating the accuracy of our methods on several shoeprint scenarios: partial prints, prints with varying levels of blurriness, prints with different amounts of wear, and prints from different shoe models. We find that models trained on one type of shoeprint yield extremely high levels of accuracy when tested on shoeprint pairs of the same scenario but fail to generalise to other scenarios. We also discover that models trained on a variety of scenarios predict almost as accurately as models trained on specific scenarios.  \nKey words: Alignment, Clustering, Forensic science, Interactive web applications, Point-set registration, Random forest  \n1. Introduction  \nFootwear outsole impressions are often found at crime scenes and can serve as powerful evidence for or against an individual’s presence at a crime scene. Outsole impressions are created either when materials picked up by a shoe (e.g. , dirt, paint, or blood) make contact with a surface or when an imprint is left by a shoe’s outsole in a substance such as mud or sand.  \nThere are three tiers of shoeprint characteristics with which forensic examiners determine whether or not a known shoeprint K from a suspect matches the questioned shoeprint Q found at the crime scene [16] . Class characteristics consist of the macro traits of a shoeprint, such as a shoe’s size, brand, make, and model. Subclass characteristics pertain to the outsole, specifically to differences in the pattern of the outsole that occur largely during manufacturing. Individual characteristics involve the changes to the outsole pattern that occur as a result of wear and tear. These traits are considered unique to the outsole of one specific shoe and are also known as randomly acquired characteristics (RACs) .  \nMethods of forensic shoeprint pattern matching have historically relied largely on visual comparisons based on guidelines proposed by Scientific Working Group for Shoeprint and Tire Tread Evidence (SWGTREAD) [6] . A human eye can reliably determine whether two shoes’ class and subclass characteristics match, but it can be difficult, even for highly  \ntrained examiners, to determine whether two prints’RACs match. Analysing RACs, however, is the most important aspect of forensic shoeprint matching since RACs are thought tobe unique to specific outsoles. As such, RACs can provide strong evidence that a shoeprint belongs to a specific shoe. Additionally, examiners currently have no way to quantify the extent to which two shoeprints match. Therefore, the focus of this work is th","cbCaiuiTdRDoDNYr","https://ap.wps.com/l/cbCaiuiTdRDoDNYr","pdf",11711242,1,20,"English","en",105,"# Introduction\n## Shoeprint evidence and characteristics\n## Limitations of visual comparison and human judgement\n## Need for quantitative automated methods\n# Proposed machine learning pipeline\n## Edge detection and 2D coordinate extraction\n## ICP alignment and similarity metrics\n## Random forest probabilistic matching\n# Generalisability evaluation","[{\"question\":\"How does the method align two shoeprints before comparing them?\",\"answer\":\"It aligns extracted 2D shoeprint coordinates using iterative closest point (ICP) after edge detection from the scans.\"},{\"question\":\"What model is used to produce a probabilistic matching score?\",\"answer\":\"A random forest is trained on similarity metrics to generate a probabilistic measurement of how likely two prints came from the same outsole.\"},{\"question\":\"What do the results show about training on lab scans versus realistic crime-scene scenarios?\",\"answer\":\"Models trained on one shoeprint scenario achieve extremely high accuracy within the same scenario but fail to generalise well to other scenarios, while multi-scenario training performs nearly as accurately across scenarios.\"}]","Improving and Evaluating Machine Learning Methods for Forensic Shoeprint Matching | PDF",1785728252,50,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"improving-and-evaluating-machine-learning-methods-for-forensic-shoeprint-matching","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/improving-and-evaluating-machine-learning-methods-for-forensic-shoeprint-matching/120111/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"How does the method align two shoeprints before comparing them?","Question",{"text":75,"@type":76},"It aligns extracted 2D shoeprint coordinates using iterative closest point (ICP) after edge detection from the scans.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What model is used to produce a probabilistic matching score?",{"text":80,"@type":76},"A random forest is trained on similarity metrics to generate a probabilistic measurement of how likely two prints came from the same outsole.",{"name":82,"@type":73,"acceptedAnswer":83},"What do the results show about training on lab scans versus realistic crime-scene scenarios?",{"text":84,"@type":76},"Models trained on one shoeprint scenario achieve extremely high accuracy within the same scenario but fail to generalise well to other scenarios, while multi-scenario training performs nearly as accurately across scenarios.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,114,119,122,126,129,133],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":29,"slug":113},6,"Technology","technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":21,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":127,"show_sort_weight":21,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":46,"category_name":135,"show_sort_weight":106,"slug":136},19,"General","general"]