New Publications within the ARTIFACT Project
We are excited to share that within the framework of the ARTIFACT project, three new research articles have been published, advancing the application of machine learning and digital technologies in water management and urban infrastructure monitoring:

Enhancing the Monitoring System for River Water Quality: Harnessing the Power of Satellite Data and Machine Learning – Velibor Ilić, Maja Turk Sekulić, Jelena Brborić, Jelena Radonić, Sonja Dmitrašinović, Milan Stojković. This study integrates Sentinel-2 satellite imagery with multiple machine learning models to enable large-scale, real-time monitoring of water temperature, electrical conductivity, and dissolved oxygen in the Danube River. LSTM and XGBoost achieved the highest predictive accuracy, while explainable AI techniques highlighted the most influential variables, offering a scalable and cost-effective solution for river basin management.
https://zenodo.org/records/16909134
Urban Flood Prediction and Mapping Using Machine Learning and Deep Learning – Jasmina Moskovljević, Anja Ranđelović, Milan Stojković, Veljko Prodanović. Presented as a poster at the 13th Urban Drainage Modelling (UDM) Conference (September 15–19, Innsbruck, Austria), this work reviews recent ML and deep learning approaches for predicting urban flood timing, extent, and damage, highlighting key input variables and model performance metrics across 112 studies.
https://zenodo.org/records/17223978
Digital Twins of Urban Drainage Systems: ML-assisted Algorithm for Processing Sensor Data – Luka Vinokić, Miloš Milašinović, Željko Vasilić, Damjan Ivetić, Milan Stojković, Veljko Prodanović. Delivered as an oral presentation at the 13th UDM Conference, this research introduces an ML-powered algorithm for automated anomaly detection and estimation of missing data in urban drainage sensor networks, tested on a synthetic dataset from the Belgrade stormwater system.
The full articles can also be found on the project website: PUBLICATIONS