HFHein Fleuren

Hein Fleuren

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hein Fleuren has a total of 73 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Explaining CNN-Based Body Measurement Estimation in Children Using Grad-RAM

    by Hezha Hassan MohammedKhan, Eric Postma, Andrew Abel, Çiçek Güven, John Burks, Liam Brown, Hein Fleuren

    Convolutional Neural Networks (CNNs) can estimate anthropometric measurements from 2D images, yet the opacity of CNN models hinders deployment in health-adjacent settings. We study explainability for regression by adapting and comparing Gradient-weighted Regression Activation Mapping (Grad-RAM) variants to four child measurements (height, weight, waist circumference, head circumference) using a DenseNet-121 backbone and a landmark-guided cropping pipeline. To the best of our knowledge, this is the first paper to apply explainable AI (XAI) methods to body measurement estimation tasks, specifically of children. We evaluate five formulations: (i) Grad-RAM (signed), (ii) a Grad-CAM reference, (iii) GradRAM (|P|), (iv) Grad-RAMP (element-wise, | · | after summation), and(v) Element-wise Grad-RAMP (|·| before summation). Quantitatively, testset MAE reaches 2.80cm (height), 1.30kg (weight), 2.30cm (waist), and 1.08cm (head), matching or improving prior ARAN baselines. Qualitatively, all Grad-RAM/RAMP variants exhibit task-aware attributions (e.g., bodywide focus for height; abdominal emphasis for waist) while also revealing reliance on contextual cues. In a targeted ablation, masking background consistently degraded accuracy across tasks, confirming sensitivity to nonanatomical signals. We use these explanations not merely as post hoc audit tools but to position it in a cycle that feeds back to data collection and model training: saliency maps diagnose dataset and preprocessing biases (e.g., ruler dependence), guide data collection (background diversity), inform augmentation (cropping and aspect-ratio handling), and motivate model choices. This cycle operationalizes Grad-RAM for improving CNN regressors in imagebased body measurement estimation task.

  • Open Access

    Inferring Body Measurements from 2D Images: A Comprehensive Review

    by Hezha Mohammedkhan, Hein Fleuren, Çíçek Güven, Eric Postma

    The prediction of anthropometric measurements from 2D body images, particularly for children, remains an under-explored area despite its potential applications in healthcare, fashion, and fitness. While pose estimation and body shape classification have garnered extensive attention, estimating body measurements and body mass index (BMI) from images presents unique challenges and opportunities. This paper provides a comprehensive review of the current methodologies, focusing on deep-learning approaches, both standalone and in combination with traditional machine-learning techniques, for inferring body measurements from facial and full-body images. We discuss the strengths and limitations of commonly used datasets, proposing the need for more inclusive and diverse collections to improve model performance. Our findings indicate that deep-learning models, especially when combined with traditional machine-learning techniques, offer the most accurate predictions. We further highlight the promise of vision transformers in advancing the field while stressing the importance of addressing model explainability. Finally, we evaluate the current state of the field, comparing recent results and focusing on the deviations from ground truth, ultimately providing recommendations for future research directions.


  • Tutorial: Addressing Underexposed Components in Operations Research Literature

    by Frans Cruijssen, Koen Peters, Hein Fleuren

    This tutorial paper discusses underrepresented practical perspectives in the operations research literature, which primarily focuses on theoretical advancements rather than documenting successful real-world applications. Drawing on experience in academia, business, and humanitarian organizations, the authors identify seven crucial—yet often overlooked—components essential for a successful analytics implementation. These components are change management, data management, model selection and validation, project management, promotion and advocacy, scoping, and software. Despite the prevalence of analytics research, these components are seldom discussed in academic literature, leaving a gap between theory and practice in analytics. We emphasize the importance of these components to achieve impactful analytics interventions and call for a more comprehensive approach to analytics research. By bridging the gap between hard analytics (mathematical modeling) and soft analytics (practical implementation), the tutorial highlights the need for scholars and practitioners to collaborate and exchange insights for more effective real-world applications of analytics. History: This paper was refereed.


  • Open Access

    A robust approach to food aid supply chains

    by Danique de Moor, Joris Wagenaar, Robert Poos, Dick Den Hertog, Hein Fleuren


  • Open Access

    A Food Bank Supply Chain Model: Tackling Bottlenecks and Optimizing Investments

    by Meike Reusken, Frans Cruijssen, Hein Fleuren

    Food banks are initiatives that redistribute food at risk of becoming waste to people that are food insecure in middle- and high-income countries. These initiatives, which have become more prevalent in recent years, face high levels of uncertainties in their supply chain. Inspired by challenges faced by the food banks in the Netherlands, this paper describes an optimization model that can assist food bank supply chains with detecting and solving bottlenecks. We aim to determine the shortages in transport, storage and food donations and propose strategic investment decisions to tackle these problems efficiently. The optimization model prioritizes investments that will have the largest positive social impact, which we define as the numberof beneficiaries that can be served by the food banks. It furthermore deals with real-world circumstances, such as decentralized organizations, data scarcity, location-specific transport and storage capacities and strong diversity in food bank operations. Application of the model in the real-life food bank supply chain of the Netherlands has resulted in an increased capacity for serving 32% more beneficiaries.

  • UN World Food Programme: Toward Zero Hunger with Analytics

    by Koen Peters, Sérgio Silva, Tim Sergio Wolter, Luis Anjos, Nina van Ettekoven, Éric Combette, Anna Melchiori, Hein Fleuren, Dick Den Hertog, Özlem Ergun

    Each year, the United Nations World Food Programme (WFP) provides food assistance to around 100 million people in more than 80 countries. Significant investments over the last decade have put planning and optimization at the forefront of tackling emergencies at WFP. A data-driven approach to managing operations has gradually become the norm and has culminated in the creation of a supply chain planning unit and savings of more than USD 150 million—enough to support two million food-insecure people for an entire year. In this paper, we describe three analytical solutions in detail: the Supply Chain Management Dashboard, which uses descriptive and predictive analytics to bring end-to-end visibility and anticipate operational issues; Optimus, which uses a mixed-integer programming model to simultaneously optimize food basket composition and supply chain planning; and DOTS, which is a data integration platform that helps WFP automate and synchronize complex data flows. Three impact studies for Iraq, South Sudan, and COVID-19 show how these tools have changed the way WFP manages its most complex operations. Through analytics, decision makers are now equipped with the insights they need to manage their operations in the best way, thereby saving and changing the lives of millions and bringing the world one step closer to zero hunger.


  • Open Access

    Optimal depot locations for humanitarian logistics service providers using robust optimization

    by V.F. Stienen, J.C. Wagenaar, D. Den Hertog, H.A. Fleuren


  • Open Access

    The Dawn of a New Common

    by Emile Aarts, Hein Fleuren, Margriet Sitskoorn, Ton Wilthagen

    On Thursday, February 27, 2020, during a live broadcast on television, Minister Bruno Bruin is handed a note saying that it has just been confirmed that a patient with the coronavirus has been identified in the Netherlands. Allegedly, it concerns a man who is placed in isolation in the Elisabeth-Tweesteden hospital in Tilburg.This is where the story of our book starts. The hospital mentioned by the minister is hardly a kilometer away from our university, Tilburg University. Things now start to develop quickly. During several weeks, the region of Tilburg becomes the “Corona Capital” of the Netherlands in terms of the number of people infected. On March 18, Minister Bruno Bruins collapses due to exhaustion during a debate in the Government’s House of Representatives. The following day, he resigns and soon after is temporarily replaced by a politician of a party that is not part of the current political coalition. Two days earlier, the country had gone into a lockdown after a historical speech of Dutch Prime Minister Mark Rutte.


  • Open Access

    Forecasting the spread of SARS-CoV-2 is inherently ambiguous given the current state of virus research

    by Melissa Koenen, Marleen Balvert, Ruud Brekelmans, Hein Fleuren, Valentijn Stienen, Joris Wagenaar

    Since the onset of the COVID-19 pandemic many researchers and health advisory institutions have focused on virus spread prediction through epidemiological models. Such models rely on virus- and disease characteristics of which most are uncertain or even unknown for SARS-CoV-2. This study addresses the validity of various assumptions using an epidemiological simulation model. The contributions of this work are twofold. First, we show that multiple scenarios all lead to realistic numbers of deaths and ICU admissions, two observable and verifiable metrics. Second, we test the sensitivity of estimates for the number of infected and immune individuals, and show that these vary strongly between scenarios. Note that the amount of variation measured in this study is merely a lower bound: epidemiological modeling contains uncertainty on more parameters than the four in this study, and including those as well would lead to an even larger set of possible scenarios. As the level of infection and immunity among the population are particularly important for policy makers, further research on virus and disease progression characteristics is essential. Until that time, epidemiological modeling studies cannot give conclusive results and should come with a careful analysis of several scenarios on virus- and disease characteristics.