Research projects in cyber-physical systems, energy efficiency, digital health, and cultural heritage preservation
Note: The projects listed below represent my participation as a staff researcher.
European & National Research Projects (Staff researcher)
myAirCoach (2015-2018) | H2020 Analysis, modelling and sensing of both physiological and environmental factors for the customized and predictive self-management of Asthma Developed a holistic mHealth personalised asthma monitoring system empowering patients to manage their own health through automated monitoring of clinical, behavioural and environmental factors.
GamECAR (2017-2018) | H2020 Gamification of EcoDriving Behaviours through Intelligent Management of dynamic car and driver information Created an innovative Serious Games platform for eco-friendly driving using OBD sensors, environmental monitoring, and cognitive modeling for personalized driver engagement.
Take-A-Breath (2018-2021) | Greek National (EPAnEK) Smart platform for Self-Management and Support of Patients with chronic respiratory diseases Developed an innovative system for personalized monitoring of COPD and asthma using ICT, gamification, bio-indicators, and intelligent devices for drug inhalation and clinical monitoring.
SmartWork (2019-2022) | H2020 Smart Age-friendly Living and Working Environment Built a worker-centric AI system for work ability sustainability supporting active and healthy ageing for older office workers through unobtrusive health/behavior monitoring and personalized interventions.
AgeingatWork (2019-2022) | H2020 Smart, Personalized and Adaptive ICT Solutions for Active, Healthy and Productive Ageing with enhanced Workability Developed ICT-based personalized systems supporting ageing workers (50+) with virtual models, computational intelligence, AR/VR tools, and an Ambient Virtual Coach.
MyOliveGroveCoach (2019-2020) | Greek National (RIS3) Analysis, modeling and multi-spectral sensing for the predictive management of Verticillium wilt in olive groves Developed intelligent system using multispectral sensing from UAVs and machine learning for early detection of Verticillium wilt in olive groves.
CPSoSaware (2020-2022) | H2020 Cross-layer cognitive optimization tools & methods for the lifecycle support of dependable CPSoS Created AI-driven tools for resource allocation in Cyber-physical Systems of Systems, evaluated in automotive and manufacturing sectors with extended reality interfaces.
EnerMan (2021-2024) | H2020 ENERgy-efficient manufacturing system MANagement Developed comprehensive framework integrating big data analytics, predictive engines, and intelligent decision support for optimizing energy consumption in manufacturing processes.
WARMEST | H2020 MSCA-RISE loW Altitude Remote sensing for the Monitoring of the state of cultural hEritage Sites Optimized maintenance procedures for cultural heritage sites (Alhambra, Santa Croce, Marzamemi) using low altitude remote sensing and Decision Support Systems.
Industry & National Research Projects (Coordinating Postdoc)
CoConstruct (2021-2026) | TUM Innovation Network Collaborative Construction Interdisciplinary research network coordinated by Prof. Kathrin Dörfler exploring novel human-machine collaboration concepts to improve productivity, resource efficiency, and quality in construction. Combines human cognitive and sensory abilities with machine learning and robotic technologies for automated design, planning, and construction.
DrawOn (2022-2025) | BayVFP Bavaria AI-based analysis of construction plans to automate project data management Coordinates development of AI and Big Data methods to analyze construction drawings using machine learning for plan type recognition, component identification, ontology-based classification, and automatic 3D model generation to optimize project data management.
FORWARD (2023-2027) | Georg Nemetschek Institute Pedestrian dynamics prediction for safe and flow-efficient building design Coordinates automation of pedestrian flow prediction in indoor/outdoor spaces using Machine Learning methods, enabling architects and engineers to interactively assess and optimize building designs for pedestrian traffic and evacuation scenarios.
AI4CADCAM (2023-2026) | Bavarian Ministry of Economic Affairs AI-based processing of CAD models for automated planning of computer-aided manufacturing Coordinates automation of CAM planning through geometry-driven generative AI, leveraging Transformer architectures, CNNs, VAEs, and Graph Neural Networks to automate tool selection, orientation, and process parameter derivation from CAD models.
BIM Automation (2023-2027) | Nemetschek SE AI-based automation of repetitive design steps in BIM modeling Coordinates development of “learning-by-modeling” approach using AI to predict optimal sequences of modeling operations, providing recommendations to reduce time and effort in BIM modeling workflows.
Publications per funding mandate
myAirCoach
2018
IEEE Access
An mHealth system for monitoring medication adherence in obstructive respiratory diseases using content based audio classification
Stavros Nousias, Aris S Lalos, Gerasimos Arvanitis, Konstantinos Moustakas, Triantafillos Tsirelis, Dimitrios Kikidis, Konstantinos Votis, and Dimitrios Tzovaras
@article{nousias2018mhealth,title={An mHealth system for monitoring medication adherence in obstructive respiratory diseases using content based audio classification},author={Nousias, Stavros and Lalos, Aris S and Arvanitis, Gerasimos and Moustakas, Konstantinos and Tsirelis, Triantafillos and Kikidis, Dimitrios and Votis, Konstantinos and Tzovaras, Dimitrios},journal={IEEE Access},volume={6},pages={11871--11882},year={2018},publisher={IEEE},doi={10.1109/ACCESS.2018.2809611},url={https://ieeexplore.ieee.org/document/8302483},projects={myAirCoach},category_1={MLHM}}
HCCAI
3-Class Prediction of Asthma Control Status Using a Gaussian Mixture Model Approach
Gerasimos Arvanitis, Otilia Kocsis, Aris S Lalos, Stavros Nousias, Konstantinos Moustakas, and Nikos Fakotakis
In Proceedings of the 10th Hellenic Conference on Artificial Intelligence, 2018
@inproceedings{arvanitis20183,title={3-Class Prediction of Asthma Control Status Using a Gaussian Mixture Model Approach},author={Arvanitis, Gerasimos and Kocsis, Otilia and Lalos, Aris S and Nousias, Stavros and Moustakas, Konstantinos and Fakotakis, Nikos},booktitle={Proceedings of the 10th Hellenic Conference on Artificial Intelligence},pages={1--2},year={2018},projects={myAirCoach},}
2017
BMC
Substance deposition assessment in obstructed pulmonary system through numerical characterization of airflow and inhaled particles attributes
@article{lalas2017substance,title={Substance deposition assessment in obstructed pulmonary system through numerical characterization of airflow and inhaled particles attributes},author={Lalas, Antonios and Nousias, Stavros and Kikidis, Dimitrios and Lalos, Aris and Arvanitis, Gerasimos and Sougles, Christos and Moustakas, Konstantinos and Votis, Konstantinos and Verbanck, Sylvia and Usmani, Omar and others},journal={BMC medical informatics and decision making},volume={17},pages={25--44},year={2017},doi={10.1186/s12911-017-0561-y},publisher={BioMed Central},category_1={BME},projects={myAirCoach},}
2016
IEEE BIBM
Numerical assessment of airflow and inhaled particles attributes in obstructed pulmonary system
Antonios Lalas, Dimitrios Kikidis, Konstantinos Votis, Dimitrios Tzovaras, Sylvia Verbanck, Stavros Nousias, A Lalos, Konstantinos Moustakas, and Omar Usmani
In 2016 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), 2016
@inproceedings{lalas2016numerical,title={Numerical assessment of airflow and inhaled particles attributes in obstructed pulmonary system},author={Lalas, Antonios and Kikidis, Dimitrios and Votis, Konstantinos and Tzovaras, Dimitrios and Verbanck, Sylvia and Nousias, Stavros and Lalos, A and Moustakas, Konstantinos and Usmani, Omar},booktitle={2016 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)},pages={606--612},year={2016},projects={myAirCoach},organization={IEEE}}
Springer
Computational modeling for simulating obstructive lung diseases based on geometry processing methods
Stavros Nousias, Aris S Lalos, and Konstantinos Moustakas
In Digital Human Modeling: Applications in Health, Safety, Ergonomics and Risk Management: 7th International Conference, DHM 2016, Held as Part of HCI International 2016, Toronto, ON, Canada, July 17-22, 2016, Proceedings 7, 2016
@inproceedings{nousias2016computational,title={Computational modeling for simulating obstructive lung diseases based on geometry processing methods},author={Nousias, Stavros and Lalos, Aris S and Moustakas, Konstantinos},booktitle={Digital Human Modeling: Applications in Health, Safety, Ergonomics and Risk Management: 7th International Conference, DHM 2016, Held as Part of HCI International 2016, Toronto, ON, Canada, July 17-22, 2016, Proceedings 7},pages={100--109},year={2016},projects={myAirCoach},organization={Springer International Publishing}}
IEEE SSCI
Monitoring asthma medication adherence through content based audio classification
Stavros Nousias, John Lakoumentas, Aris Lalos, Dimitrios Kikidis, Konstantinos Moustakas, Konstantinos Votis, and Dimitrios Tzovaras
In 2016 IEEE symposium series on computational intelligence (SSCI), 2016
@inproceedings{nousias2016monitoring,title={Monitoring asthma medication adherence through content based audio classification},author={Nousias, Stavros and Lakoumentas, John and Lalos, Aris and Kikidis, Dimitrios and Moustakas, Konstantinos and Votis, Konstantinos and Tzovaras, Dimitrios},booktitle={2016 IEEE symposium series on computational intelligence (SSCI)},pages={1--5},year={2016},projects={myAirCoach},organization={IEEE}}
GamECAR
2019
ENTSC
Exploiting gamification to improve eco-driving behaviour: The GamECAR approach
Stavros Nousias, Christos Tselios, Dimitris Bitzas, Dimitris Amaxilatis, Javier Montesa, Aris S Lalos, Konstantinos Moustakas, and Ioannis Chatzigiannakis
Electronic Notes in Theoretical Computer Science, 2019
@article{nousias2019exploiting,title={Exploiting gamification to improve eco-driving behaviour: The GamECAR approach},author={Nousias, Stavros and Tselios, Christos and Bitzas, Dimitris and Amaxilatis, Dimitris and Montesa, Javier and Lalos, Aris S and Moustakas, Konstantinos and Chatzigiannakis, Ioannis},journal={Electronic Notes in Theoretical Computer Science},volume={343},pages={103--116},year={2019},publisher={ACM},projects={GamECAR},}
EPS
Gamification of EcoDriving Behaviours through Intelligent Management of dynamic car and driver information
Nousias Stavros, Aris S Lalos, Christos Tselios, Dimitris Bitzas, Dimitris Amaxilatis, Ioannis Chatzigiannakis, Arvanitis Gerasimos, and Konstantinos Moustakas
OPPORTUNITIES AND CHALLENGES for European Projects (EPS Portugal 2017/2018 2017), 2019
@article{stavros2019gamification,title={Gamification of EcoDriving Behaviours through Intelligent Management of dynamic car and driver information},author={Stavros, Nousias and Lalos, Aris S and Tselios, Christos and Bitzas, Dimitris and Amaxilatis, Dimitris and Chatzigiannakis, Ioannis and Gerasimos, Arvanitis and Moustakas, Konstantinos},journal={OPPORTUNITIES AND CHALLENGES for European Projects (EPS Portugal 2017/2018 2017)},pages={100--123},year={2019},projects={GamECAR},}
ENTSC
Exploiting gamification to improve eco-driving behaviour: The GamECAR approach
Stavros Nousias, Christos Tselios, Dimitris Bitzas, Dimitris Amaxilatis, Javier Montesa, Aris S Lalos, Konstantinos Moustakas, and Ioannis Chatzigiannakis
Electronic Notes in Theoretical Computer Science, 2019
@inproceedings{tselios2019enhancing,title={Enhancing an eco-driving gamification platform through wearable and vehicle sensor data integration},author={Tselios, Christos and Nousias, Stavros and Bitzas, Dimitris and Amaxilatis, Dimitrios and Akrivopoulos, Orestis and Lalos, Aris S and Moustakas, Konstantinos and Chatzigiannakis, Ioannis},booktitle={Ambient Intelligence: 15th European Conference, AmI 2019, Rome, Italy, November 13--15, 2019, Proceedings 15},pages={344--349},year={2019},organization={Springer International Publishing},doi={10.1007/978-3-030-34255-5_26},category_1={Gamification},projects={GamECAR},}
2018
IEEE CAMAD
Uncertainty management for wearable iot wristband sensors using laplacian-based matrix completion
Stavros Nousias, Christos Tselios, Dimitris Bitzas, Aris S Lalos, Konstantinos Moustakas, and Ioannis Chatzigiannakis
In 2018 IEEE 23rd International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD), 2018
Contemporary sensing devices provide reliable mechanisms for continuous process monitoring, accommodating use cases related to mHealth and smart mobility, by generating real-time data streams of numerous physiological and vital parameters. Such data streams can be later utilized by machine learning algorithms and decision support systems to predict critical clinical states and motivate users to adopt behaviours that improve the quality of their life and the society as a whole. However, in many cases, even when deployed over highly sophisticated, cutting-edge network infrastructure and deployment paradigms, data may exhibit missing values and non-uniformities due to various reasons, including device malfunction, deliberate data reduction for efficient processing, or data loss due to sensing and communication failures. This work proposes a novel approach to deal with missing entries in heart rate measurements. Benefiting from the low-rank property of the generated data matrices and the proximity of neighbouring measurements, we provide a novel method that combines classical matrix completion approaches with weighted Laplacian interpolation offering high reconstruction accuracy at fast execution times. Extensive evaluation studies carried out with real measurements show that the proposed methods could be effectively deployed by modern wristband-cloud computing systems increasing the robustness, the reliability and the energy efficiency of these systems.
@inproceedings{nousias2018uncertainty,title={Uncertainty management for wearable iot wristband sensors using laplacian-based matrix completion},author={Nousias, Stavros and Tselios, Christos and Bitzas, Dimitris and Lalos, Aris S and Moustakas, Konstantinos and Chatzigiannakis, Ioannis},booktitle={2018 IEEE 23rd International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD)},pages={1--6},year={2018},organization={IEEE},doi={10.1109/CAMAD.2018.8515001},category_1={MCDI},projects={GamECAR},}
IEEE PerCom
Managing nonuniformities and uncertainties in vehicle-oriented sensor data over next generation networks
Stavros Nousias, Christos Tselios, Olivier Orfila, Samantha Jamson, Pablo Mejuto, Dimitrios Amaxilatis, Orestis Akrivopoulos, Ioannis Chatzigiannakis, Aris S Lalos, and Konstantinos Moustakas
In 2018 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops), 2018
Detailed and accurate vehicle-oriented sensor data is considered fundamental for efficient vehicle-to-everything V2X communication applications, especially in the upcoming highly heterogeneous, brisk and agile 5G networking era. Information retrieval, transfer and manipulation in real-time offers a small margin for erratic behavior, regardless of its root cause. This paper presents a method for managing nonuniformities and uncertainties found on datasets, based on an elaborate Matrix Completion technique, with superior performance in three distinct cases of vehicle-related sensor data, collected under real driving conditions. Our approach appears capable of handling sensing and communication irregularities, minimizing at the same time the storage and transmission requirements of Multi-access Edge Computing applications.
@inproceedings{nousias2018managing,title={Managing nonuniformities and uncertainties in vehicle-oriented sensor data over next generation networks},author={Nousias, Stavros and Tselios, Christos and Orfila, Olivier and Jamson, Samantha and Mejuto, Pablo and Amaxilatis, Dimitrios and Akrivopoulos, Orestis and Chatzigiannakis, Ioannis and Lalos, Aris S and Moustakas, Konstantinos},booktitle={2018 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)},pages={272--277},year={2018},organization={IEEE},doi={10.1109/PERCOMW.2018.8480342},category_1={MCDI},projects={GamECAR}}
SmartWork
2019
IEEE BIA
Coping with missing data in an unobtrusive monitoring system for office workers
Stavros Nousias, Giwrgos Papoulias, Otilia Kocsis, Miriam Cabrita, Aris S Lalos, and Konstantinos Moustakas
In 2019 International Conference on Biomedical Innovations and Applications (BIA), 2019
@inproceedings{nousias2019coping,title={Coping with missing data in an unobtrusive monitoring system for office workers},author={Nousias, Stavros and Papoulias, Giwrgos and Kocsis, Otilia and Cabrita, Miriam and Lalos, Aris S and Moustakas, Konstantinos},booktitle={2019 International Conference on Biomedical Innovations and Applications (BIA)},pages={1--4},year={2019},organization={IEEE},doi={10.1109/BIA48344.2019.8967465},projects={SmartWork},category_1={MCDI}}
Ageing@Work
2019
IEEE INDIN
Fast mesh denoising with data driven normal filtering using deep autoencoders
Stavros Nousias, Gerasimos Arvanitis, Aris S Lalos, and Konstantinos Moustakas
In 2019 IEEE 17th International Conference on Industrial Informatics (INDIN), 2019
@inproceedings{nousias2019fast,title={Fast mesh denoising with data driven normal filtering using deep autoencoders},author={Nousias, Stavros and Arvanitis, Gerasimos and Lalos, Aris S and Moustakas, Konstantinos},booktitle={2019 IEEE 17th International Conference on Industrial Informatics (INDIN)},volume={1},pages={260--263},year={2019},organization={IEEE},projects={Ageing@Work},}
I3T
2019
IEEE SA
Sparse modeling and optimization tools for energy efficient and reliable IoT
Stavros Nousias, Aris S Lalos, Athanasios Kalogeras, Christos Alexakos, Christos Koulamas, and Konstantinos Moustakas
In 2019 First International Conference on Societal Automation (SA), 2019
@inproceedings{nousias2019sparse,title={Sparse modeling and optimization tools for energy efficient and reliable IoT},author={Nousias, Stavros and Lalos, Aris S and Kalogeras, Athanasios and Alexakos, Christos and Koulamas, Christos and Moustakas, Konstantinos},booktitle={2019 First International Conference on Societal Automation (SA)},pages={1--4},year={2019},organization={IEEE},projects={I3T},doi={10.1109/SA47457.2019.8938029},category_1={MCDI}}
DEEP-EVIoT
2019
IEEE ETFA
Assessment of medication adherence in respiratory diseases through deep sparse convolutional coding
Vaggelis Ntalianis, Stavros Nousias, Aris S Lalos, Michael Birbas, Nikolaos Tsafas, and Konstantinos Moustakas
In 2019 24th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA), 2019
@inproceedings{ntalianis2019assessment,title={Assessment of medication adherence in respiratory diseases through deep sparse convolutional coding},author={Ntalianis, Vaggelis and Nousias, Stavros and Lalos, Aris S and Birbas, Michael and Tsafas, Nikolaos and Moustakas, Konstantinos},booktitle={2019 24th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA)},pages={1657--1660},year={2019},projects={DEEP-EVIoT},organization={IEEE},}
MyOliveGroveCoach
2020
IEEE INDIN
Efficient automated U-Net based tree crown delineation using UAV multi-spectral imagery on embedded devices
Kostas Blekos, Stavros Nousias, and Aris S Lalos
In 2020 IEEE 18th International Conference on Industrial Informatics (INDIN), 2020
Delineation approaches provide significant benefits to various domains, including agriculture, environmental and natural disasters monitoring. Most of the work in the literature utilize traditional segmentation methods that require a large amount of computational and storage resources. Deep learning has transformed computer vision and dramatically improved machine translation, though it requires massive dataset for training and significant resources for inference. More importantly, energy-efficient embedded vision hardware delivering real-time and robust performance is crucial in the aforementioned application. In this work, we propose a U-Net based tree delineation method, which is effectively trained using multi-spectral imagery but can then delineate single-spectrum images. The deep architecture that also performs localization, i.e., a class label corresponds to each pixel, has been successfully used to allow training with a small set of segmented images. The ground truth data were generated using traditional image denoising and segmentation approaches. To be able to execute the proposed DNN efficiently in embedded platforms designed for deep learning approaches, we employ traditional model compression and acceleration methods. Extensive evaluation studies using data collected from UAVs equipped with multi-spectral cameras demonstrate the effectiveness of the proposed methods in terms of delineation accuracy and execution efficiency.
@inproceedings{blekos2020efficient,title={Efficient automated U-Net based tree crown delineation using UAV multi-spectral imagery on embedded devices},author={Blekos, Kostas and Nousias, Stavros and Lalos, Aris S},booktitle={2020 IEEE 18th International Conference on Industrial Informatics (INDIN)},volume={1},pages={541--546},year={2020},organization={IEEE},category_1={MCA},projects={MyOlivGroveCoach},}
CPSoSaware
2023
IEEE Access
Accelerating deep neural networks for efficient scene understanding in multi-modal automotive applications
Stavros Nousias, Erion-Vasilis Pikoulis, Christos Mavrokefalidis, and Aris S Lalos
Environment perception constitutes one of the most critical operations performed by semi- and fully- autonomous vehicles. In recent years, Deep Neural Networks (DNNs) have become the standard tool for perception solutions owing to their impressive capabilities in analyzing and modelling complex and dynamic scenes, from (often multi-modal) sensory inputs. However, the well-established performance of DNNs comes at the cost of increased time and storage complexity, which may become problematic in automotive perception systems due to the requirement for a short prediction horizon (as in many cases inference must be performed in real-time) and the limited computational, storage, and energy resources of mobile systems. A common way of addressing this problem is to transform the original large pre-trained networks into new smaller models, by utilizing Model Compression and Acceleration (MCA) techniques, improving both their storage and execution efficiency. Within the MCA framework, in this paper, we investigate the application of two state-of-the-art weight-sharing MCA techniques, namely a Vector Quantization (VQ) and a Dictionary Learning (DL) one, as well as two novel extensions, towards the acceleration and compression of widely used DNNs for 2D and 3D object-detection in automotive applications. Apart from the individual (uni-modal) networks, we also present and evaluate a multi-modal late-fusion algorithm for combining the detection results of the 2D and 3D detectors. Our evaluation studies are carried out on the KITTI Dataset. The obtained results lend themselves to a twofold interpretation. On the one hand, they showcase the significant acceleration and compression gains that can be achieved via the application of weight sharing on the selected DNN detectors, with limited accuracy loss, as well as highlight the performance differences between the two utilized weight-sharing approaches. On the other, they demonstrate the substantial boost in detection performance obtained by combining the outcome of the two unimodal individual detectors, using the proposed late-fusion-based multi-modal approach. Indeed, as our experiments have shown, pairing the high-performance DL-based MCA technique with the loss-mitigating effect of the multi-modal fusion approach, leads to highly accelerated models (up to approximately 2.5× and 6× for the 2D and 3D detectors, respectively) with the performance loss of the fused results ranging in most cases within single-digits figures (as low as around 1% for the class “cars”).
@article{nousias2023accelerating,title={Accelerating deep neural networks for efficient scene understanding in multi-modal automotive applications},author={Nousias, Stavros and Pikoulis, Erion-Vasilis and Mavrokefalidis, Christos and Lalos, Aris S},journal={IEEE Access},doi={10.1109/ACCESS.2023.3258400},url={https://ieeexplore.ieee.org/document/10073550},volume={11},pages={28208--28221},year={2023},publisher={IEEE},category_2={MCA},projects={CPSoSaware},}
2022
Springer
A new clustering-based technique for the acceleration of deep convolutional networks
Erion Vasilis Pikoulis, Christos Mavrokefalidis, Stavros Nousias, and Aris S Lalos
Deep learning and especially the use of Deep Neural Networks (DNNs) provide impressive results in various regression and classification tasks. However, to achieve these results, there is a high demand for computing and storing resources. This becomes problematic when, for instance, real-time mobile applications are considered, in which the involved (embedded) devices have limited resources. A common way of addressing this problem is to transform the original large pre-trained networks into new smaller models, by utilizing Model Compression and Acceleration (MCA) techniques. Within the MCA framework, we propose a clustering-based approach that is able to increase the number of employed centroids/representatives, while, at the same time, having an acceleration gain compared to conventional, k-means-based approaches. This is achieved by imposing a special structure to the employed representatives, which is enabled by the particularities of the problem at hand. Moreover, the theoretical acceleration gains are presented and the key system hyper-parameters that affect that gain are identified. Extensive evaluation studies carried out using various state-of-the-art DNN models trained in image classification and object detection validate the effectiveness of the proposed method in MCA tasks.
@article{pikoulis2022new,title={A new clustering-based technique for the acceleration of deep convolutional networks},author={Pikoulis, Erion Vasilis and Mavrokefalidis, Christos and Nousias, Stavros and Lalos, Aris S},journal={Springer,Deep Learning Applications, Volume 3},pages={123--150},year={2022},url={https://link.springer.com/chapter/10.1007/978-981-16-3357-7_5},publisher={Springer Singapore},doi={10.1007/978-981-16-3357-7_5},category_1={MCA},projects={CPSoSaware},}
2021
IEEE ICPS
Accelerating deep neural networks for efficient scene understanding in automotive cyber-physical systems
Stavros Nousias, Erion Vasilis Pikoulis, Christos Mavrokefalidis, and Aris S Lalos
In 2021 4th IEEE International Conference on Industrial Cyber-Physical Systems (ICPS), 2021
Automotive Cyber-Physical Systems (ACPS) have attracted a significant amount of interest in the past few decades, while one of the most critical operations in these systems is the perception of the environment. Deep learning and, especially, the use of Deep Neural Networks (DNNs) provides impressive results in analyzing and understanding complex and dynamic scenes from visual data. The prediction horizons for those perception systems are very short and inference must often be performed in real time, stressing the need of transforming the original large pre-trained networks into new smaller models, by utilizing Model Compression and Acceleration (MCA) techniques. Our goal in this work is to investigate best practices for appropriately applying novel weight sharing techniques, optimizing the available variables and the training procedures towards the significant acceleration of widely adopted DNNs. Extensive evaluation studies carried out using various state-of-the-art DNN models in object detection and tracking experiments, provide details about the type of errors that manifest after the application of weight sharing techniques, resulting in significant acceleration gains with negligible accuracy losses.
@inproceedings{nousias2021accelerating,title={Accelerating deep neural networks for efficient scene understanding in automotive cyber-physical systems},author={Nousias, Stavros and Pikoulis, Erion Vasilis and Mavrokefalidis, Christos and Lalos, Aris S},booktitle={2021 4th IEEE International Conference on Industrial Cyber-Physical Systems (ICPS)},pages={63--69},year={2021},organization={IEEE},doi={10.1109/ICPS49255.2021.9468126},category_1={MCA},projects={CPSoSaware}}
IEEE VLSI-SoC
Accelerating 3D scene analysis for autonomous driving on embedded AI computing platforms
Stavros Nousias, Erion-Vasilis Pikoulis, Christos Mavrokefalidis, Aris S Lalos, and Konstantinos Moustakas
In 2021 IFIP/IEEE 29th International Conference on Very Large Scale Integration (VLSI-SoC), 2021
@inproceedings{nousias2021acceleratinh,title={Accelerating 3D scene analysis for autonomous driving on embedded AI computing platforms},author={Nousias, Stavros and Pikoulis, Erion-Vasilis and Mavrokefalidis, Christos and Lalos, Aris S and Moustakas, Konstantinos},booktitle={2021 IFIP/IEEE 29th International Conference on Very Large Scale Integration (VLSI-SoC)},pages={1--6},year={2021},organization={IEEE},projects={CPSoSaware},category_2={MCA}}
WARMEST
WARMEST—loW Altitude Remote sensing for the Monitoring of the state of cultural hEritage Sites: building an inTegrated model for maintenance—is an H2020 Marie Curie Research and Innovation Staff Mobility Project (H2020 Marie Skłodowska-Curie Actions, RISE 2017).
The WARMEST project is carried out by an international and multi-sectorial consortium formed by more than 30 researchers belonging to 3 academic institutions, 5 private companies, and 3 associations from 4 different countries.
In close cooperation with the consortium, 3 heritage sites are involved in the research: Patronato de la Alhambra y Generalife (Granada, Spain), Marzamemi Underwater Museum (Sicily, Italy), and Opera Santa Croce (Florence, Italy).
WARMEST’s strategic goal is to optimize maintenance procedures in cultural and natural heritage sites through the introduction of new technologies to collect data and new tools to analyse it, creating a novel Decision Support System. This software enables stakeholders to set, on sound economic and technical bases, the most suitable maintenance procedures taking into account current and future scenarios. It thus helps prepare sound and cost-effective maintenance plans, given that maintenance of monuments and sites is a challenging management and engineering activity due to the large variety of loads (different types of sites and actions), strict regulations to preserve natural, archaeological, and historical values, and continuous operation throughout the year.
2023
ACM TMM
Deep saliency mapping for 3D meshes and applications
Stavros Nousias, Gerasimos Arvanitis, Aris Lalos, and Konstantinos Moustakas
ACM Transactions on Multimedia Computing, Communications and Applications, 2023
@article{nousias2023deep,title={Deep saliency mapping for 3D meshes and applications},author={Nousias, Stavros and Arvanitis, Gerasimos and Lalos, Aris and Moustakas, Konstantinos},journal={ACM Transactions on Multimedia Computing, Communications and Applications},volume={19},number={2},pages={1--22},year={2023},doi={10.1145/3550073},url={https://dl.acm.org/doi/10.1145/3550073},publisher={ACM New York, NY},category_1={GDL},projects={WARMEST},}
2022
IEEE ICPS
Coarse-to-fine defect detection of heritage 3D models using a CNN learning approach
Gerasimos Arvanitis, Stavros Nousias, Aris S Lalos, and Konstantinos Moustakas
In 2022 IEEE 5th International Conference on Industrial Cyber-Physical Systems (ICPS), 2022
@inproceedings{arvanitis2022coarse,title={Coarse-to-fine defect detection of heritage 3D models using a CNN learning approach},author={Arvanitis, Gerasimos and Nousias, Stavros and Lalos, Aris S and Moustakas, Konstantinos},booktitle={2022 IEEE 5th International Conference on Industrial Cyber-Physical Systems (ICPS)},pages={1--6},doi={10.1109/ICPS56567.2022.9899335},url={https://ieeexplore.ieee.org/abstract/document/9816869},year={2022},organization={IEEE},category_1={HTG},projects={WARMEST},category_2={GDL}}
2020
IEEE TII
Fast mesh denoising with data driven normal filtering using deep variational autoencoders
Stavros Nousias, Gerasimos Arvanitis, Aris S Lalos, and Konstantinos Moustakas
@article{nousias2020fast,title={Fast mesh denoising with data driven normal filtering using deep variational autoencoders},author={Nousias, Stavros and Arvanitis, Gerasimos and Lalos, Aris S and Moustakas, Konstantinos},journal={IEEE Transactions on Industrial Informatics},volume={17},number={2},doi={10.1109/TII.2020.3000491},url={https://ieeexplore.ieee.org/document/9110709},pages={980--990},year={2020},publisher={IEEE},category_1={GDL},projects={WARMEST},}
IEEE Access
A saliency aware CNN-based 3D model simplification and compression framework for remote inspection of heritage sites
Stavros Nousias, Gerasimos Arvanitis, Aris S Lalos, George Pavlidis, Christos Koulamas, Athanasios Kalogeras, and Konstantinos Moustakas
@article{nousias2020saliency,title={A saliency aware CNN-based 3D model simplification and compression framework for remote inspection of heritage sites},author={Nousias, Stavros and Arvanitis, Gerasimos and Lalos, Aris S and Pavlidis, George and Koulamas, Christos and Kalogeras, Athanasios and Moustakas, Konstantinos},journal={IEEE Access},volume={8},doi={10.1109/ACCESS.2020.3023167},url={https://ieeexplore.ieee.org/document/9193917},pages={169982--170001},year={2020},publisher={IEEE},category_1={GDL},projects={WARMEST},}
2019
IEEE ICME
Mesh saliency detection using convolutional neural networks
Stavros Nousias, Gerasimos Arvanitis, Aris S Lalos, and Konstantinos Moustakas
In 2020 IEEE international conference on multimedia and expo (ICME), 2019
@inproceedings{nousias2020mesh,title={Mesh saliency detection using convolutional neural networks},author={Nousias, Stavros and Arvanitis, Gerasimos and Lalos, Aris S and Moustakas, Konstantinos},booktitle={2020 IEEE international conference on multimedia and expo (ICME)},pages={1--6},year={2019},doi={10.1109/ICME46284.2020.9102796},organization={IEEE},category_1={GDL},projects={WARMEST},}
Take-A-Breath
Take-A-Breath: Smart Platform for Self-management and Support of Patients with Chronic Respiratory Diseases
2023
IEEE Access
AI sound recognition on asthma medication adherence: Evaluation with the RDA benchmark suite
Dimitris Nikos Fakotakis, Stavros Nousias, Gerasimos Arvanitis, Evangelia I Zacharaki, and Konstantinos Moustakas
Asthma is a common, usually long-term respiratory disease with negative impact on global society and economy. Treatment involves using medical devices (inhalers) that distribute medication to the airways and its efficiency depends on the precision of the inhalation technique. There is a clinical need for objective methods to assess the inhalation technique, during clinical consultation. Integrated health monitoring systems, equipped with sensors, enable the recognition of drug actuation, embedded with sound signal detection, analysis and identification from intelligent structures, that could provide powerful tools for reliable content management. Health monitoring systems equipped with sensors, embedded with sound signal detection, enable the recognition of drug actuation and could be used for effective audio content analysis. This paper revisits sound pattern recognition with machine learning techniques for asthma medication adherence assessment and presents the Respiratory and Drug Actuation (RDA) Suite (https://gitlab.com/vvr/monitoring-medication-adherence/rda-benchmark) for benchmarking and further research. The RDA Suite includes a set of tools for audio processing, feature extraction and classification procedures and is provided along with a dataset, consisting of respiratory and drug actuation sounds. The classification models in RDA are implemented based on conventional and advanced machine learning and deep networks’ architectures. This study provides a comparative evaluation of the implemented approaches, examines potential improvements and discusses on challenges and future tendencies.
@article{fakotakis2023ai,title={AI sound recognition on asthma medication adherence: Evaluation with the RDA benchmark suite},author={Fakotakis, Dimitris Nikos and Nousias, Stavros and Arvanitis, Gerasimos and Zacharaki, Evangelia I and Moustakas, Konstantinos},journal={IEEE Access},volume={11},pages={13810--13829},year={2023},doi={10.1109/ACCESS.2023.3243547},url={https://ieeexplore.ieee.org/document/10040762},publisher={IEEE},category_1={MLHM},projects={Take-A-Breath},}
2020
PLoS ONE
AVATREE: An open-source computational modelling framework modelling Anatomically Valid Airway TREE conformations
Stavros Nousias, Evangelia I. Zacharaki, and Konstantinos Moustakas
@article{nousias2020avatree,title={AVATREE: An open-source computational modelling framework modelling Anatomically Valid Airway TREE conformations},author={Nousias, Stavros and Zacharaki, Evangelia I. and Moustakas, Konstantinos},journal={PLoS ONE},volume={15},number={4},doi={10.1371/journal.pone.0230259},url={https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0230259},pages={e0230259},year={2020},projects={Take-A-Breath},category_1={BME}}
Sensors
Deep CNN sparse coding for real time inhaler sounds classification
Vaggelis Ntalianis, Nikos Dimitris Fakotakis, Stavros Nousias, Aris S Lalos, Michael Birbas, Evangelia I Zacharaki, and Konstantinos Moustakas
@article{ntalianis2020deep,title={Deep CNN sparse coding for real time inhaler sounds classification},author={Ntalianis, Vaggelis and Fakotakis, Nikos Dimitris and Nousias, Stavros and Lalos, Aris S and Birbas, Michael and Zacharaki, Evangelia I and Moustakas, Konstantinos},journal={Sensors},volume={20},number={8},pages={2363},year={2020},publisher={MDPI},projects={Take-A-Breath},}
2019
IEEE BIBE
Recognition of breathing activity and medication adherence using LSTM neural networks
Dionysis Pettas, Stavros Nousias, Evangelia I Zacharaki, and Konstantinos Moustakas
In 2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE), 2019
@inproceedings{pettas2019recognition,title={Recognition of breathing activity and medication adherence using LSTM neural networks},author={Pettas, Dionysis and Nousias, Stavros and Zacharaki, Evangelia I and Moustakas, Konstantinos},booktitle={2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)},pages={941--946},year={2019},organization={IEEE},doi={10.1109/BIBE.2019.00176},url={https://ieeexplore.ieee.org/document/8941960},category_1={MLHM},projects={Take-A-Breath},}
DrawOn
2025
ASCE JCCE
VectorGraphNET: Graph Attention Networks for Accurate Segmentation of Complex Technical Drawings
Andrea Carrara, Stavros Nousias, and André Borrmann
This study introduces a new approach to extracting vector data from technical drawings in Portable Document Format (PDF) format and analyzing them semantically by employing graph attention networks. The proposed method involves converting PDF files into Scalable Vector Graphics (SVG) format and creating a feature-rich graph representation, which captures the relationships between vector entities using geometrical information. We then apply a graph attention transformer with hierarchical label definition to achieve accurate line-level semantic segmentation. Our approach is evaluated on two data sets, including the public FloorplanCAD data set, which achieves state-of-the-art results on weighted F1-score (89%), surpassing existing methods. Our vector-based method offers a scalable large-scale technical drawing analysis solution, requiring significantly less graphical processing unit (GPU) power than does current state-of-the-art techniques. Our method performs better semantic segmentation tasks, effectively extracting meaningful information from technical drawings. This enables reliable line-level classification, even for complex drawings such as architectural floorplans with overlapping structural and annotation layers, and opens up new applications, such as automated drawing processing. Additionally, our method can improve existing workflows in the Architecture, Engineering, and Construction (AEC) industry, including automated building information modeling (BIM) and construction planning. Technical drawings are essential for communication in the AEC industries; however, manual analysis is often time-consuming and error prone. This study presents an efficient method for automating the segmentation and analysis of technical drawings using graph-based machine-learning techniques. Converting vector data from commonly used PDF or computer-aided design (CAD) files into feature-rich graph representations enables precise and scalable classification of drawing elements, such as walls, windows, symbols, and annotations. A key advantage of this method is its efficiency, requiring limited computational resources. Unlike traditional machine learning models that rely on computationally intensive pixel-based approaches, this framework uses lightweight graph neural networks, making it accessible for professionals without access to high-performance hardware. This efficiency allows it to process large and complex technical drawings on standard GPUs or even in cloud-based environments, ensuring broader adoption across diverse organizational scales. In practice, this approach reduces project documentation time, improves accuracy, and streamlines workflows such as BIM. It is particularly beneficial for large-scale projects with intricate technical drawings, for which traditional methods struggle to maintain precision. This research demonstrates the transformative potential of AI-driven tools to make data-driven construction and planning more efficient and accessible to a wider range of practitioners.
2024
ASCE
Employing graph neural networks for construction drawing content recognition
Andrea Carrara, Stavros Nousias, and A Borrmann
In i3CE 2024: 2024 ASCE International Conference on Computing in Civil Engineering, 2024
@inproceedings{carrara2024employing,title={Employing graph neural networks for construction drawing content recognition},author={Carrara, Andrea and Nousias, Stavros and Borrmann, A},booktitle={i3CE 2024: 2024 ASCE International Conference on Computing in Civil Engineering},year={2024},projects={DrawOn},category_1={AIBE}}
PredictBIM
PredictBIM: AI-based automation of repetitive design steps in BIM modeling
2026
ASCE JCCE
Text2BIM: Generating Building Models Using a Large Language Model-based Multi-Agent Framework
Changyu Du, Sebastian Esser, Stavros Nousias, and André Borrmann
@article{du2026text2bim,title={Text2BIM: Generating Building Models Using a Large Language Model-based Multi-Agent Framework},author={Du, Changyu and Esser, Sebastian and Nousias, Stavros and Borrmann, Andr{\'e}},journal={Journal of Computing in Civil Engineering},volume={40},number={2},publisher={American Society of Civil Engineers},pages={04025142},year={2026},doi={10.1061/JCCEE5.CPENG-6386},url={https://ascelibrary.org/doi/full/10.1061/JCCEE5.CPENG-6386},category_1={AIBE},projects={PredictBIM},}
2025
ICML WCUA
BIMgent: Towards Autonomous Building Modeling via Computer-use Agents
Zihan Deng, Changyu Du, Stavros Nousias, and André Borrmann
In ICML 2025 Workshop on Computer-use Agents, Jun 2025
@inproceedings{deng2025bimgent,title={BIMgent: Towards Autonomous Building Modeling via Computer-use Agents},author={Deng, Zihan and Du, Changyu and Nousias, Stavros and Borrmann, Andr{\'e}},booktitle={ICML 2025 Workshop on Computer-use Agents},doi={10.48550/arXiv.2506.07217},month=jun,year={2025},url={https://openreview.net/forum?id=vQCFwXAl2A},projects={PredictBIM},category_1={AIBE}}
Predictive modeling: BIM command recommendation based on large-scale usage logs
Changyu Du, Zihan Deng, Stavros Nousias, and André Borrmann
The adoption of Building Information Modeling (BIM) and model-based design within the Architecture, Engineering, and Construction (AEC) industry has been hindered by the perception that using BIM authoring tools demands more effort than conventional 2D drafting. To enhance design efficiency, this paper proposes a BIM command recommendation framework that predicts the optimal next actions in real-time based on users’ historical interactions. We propose a comprehensive filtering and enhancement method for large-scale raw BIM log data and introduce a novel command recommendation model. Our model builds upon the state-of-the-art Transformer backbones originally developed for large language models (LLMs), incorporating a custom feature fusion module, dedicated loss function, and targeted learning strategy. In a case study, the proposed method is applied to over 32 billion rows of real-world log data collected globally from the BIM authoring software Vectorworks. Experimental results demonstrate that our method can learn universal and generalizable modeling patterns from anonymous user interaction sequences across different countries, disciplines, and projects. When generating recommendations for the next command, our approach achieves a Recall@10 of approximately 84%. The code is available at: https://github.com/dcy0577/BIM-Command-Recommendation.git.
FORWARD
FORWARD: Pedestrian dynamics prediction for safe and flow-efficient building design
2026
Machine learning in pedestrian and evacuation dynamics for the built environment: A systematic literature review
Patrick Berggold, Ana Čukarska, Stavros Nousias, Felix Dietrich, and André Borrmann
@article{berggold2026machine,title={Machine learning in pedestrian and evacuation dynamics for the built environment: A systematic literature review},author={Berggold, Patrick and {\v{C}}ukarska, Ana and Nousias, Stavros and Dietrich, Felix and Borrmann, Andr{\'e}},journal={Safety Science},volume={198},pages={107143},year={2026},projects={FORWARD},doi={10.1016/j.ssci.2026.107143},category_1={AIBE},url={https://www.sciencedirect.com/science/article/pii/S0925753526000342},publisher={Elsevier}}
2024
ASCE
Integrating pedestrian simulation into early building design: A deep learning approach for trajectory prediction
Patrick Berggold, Stavros Nousias, and Andre Borrmann
In i3CE 2024: 2024 ASCE International Conference on Computing in Civil Engineering, 2024
@inproceedings{berggold2024integrating,title={Integrating pedestrian simulation into early building design: A deep learning approach for trajectory prediction},author={Berggold, Patrick and Nousias, Stavros and Borrmann, Andre},booktitle={i3CE 2024: 2024 ASCE International Conference on Computing in Civil Engineering},year={2024},projects={FORWARD},category_1={AIBE}}
@inproceedings{nousias2025hodgeformer,title={HodgeFormer: Transformers for Learnable Operators on Triangular Meshes through Data-Driven Hodge Matrices},author={Nousias, Akis and Nousias, Stavros},booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},year={2026},note={To be presented},doi={10.48550/arXiv.2509.01839},url={https://hodgeformer.github.io/},projects={CoConstruct},category_1={GDL}}
2024
arXiv
Coordinating robotized construction using advanced robotic simulation: The case of collaborative brick wall assembly
Mohammad Reza Kolani, Stavros Nousias, and André Borrmann
@article{kolani2024coordinating,title={Coordinating robotized construction using advanced robotic simulation: The case of collaborative brick wall assembly},author={Kolani, Mohammad Reza and Nousias, Stavros and Borrmann, Andr{\'e}},journal={arXiv preprint arXiv:2405.17171},year={2024},projects={CoConstruct},category_1={AIBE}}
AI4CADCAM
AI4CADCAM: AI-based processing of CAD models for automated planning of computer-aided manufacturing
AICHECK
AICHECK: AI-enabled code compliance checking in early building design
2025
Elsevier
Design Healing framework for automated code compliance
@article{wu2025design,title={Design Healing framework for automated code compliance},author={Wu, Jiabin and Nousias, Stavros and Borrmann, Andr{\'e}},journal={Elsevier,Automation in Construction},volume={171},pages={106004},doi={10.1016/j.autcon.2025.106004},url={https://www.sciencedirect.com/science/article/pii/S0926580525000445},year={2025},publisher={Elsevier},projects={AICHECK},category_1={AIBE}}
