137 research outputs found
EXPERIMEDIA: Technology enablers for a future media Internet testing facility
No full textCreating innovative Future Media Internet (FMI) products and services is a complex endeavour requiring consideration of socio-technical factors and an increasingly diverse technology landscape. Accelerating time to market requires the availability of technology enablers adapted to local contexts and integrated together to create added value patterns of use. In this paper we present a set of such technology enablers used within a FMI testing facility. We describe how each enabler has been constructed to support the lifecycle of different classes of content and integrated to provide coherent and representative aggregations of content expected in FMI applications and services. A set of lessons learnt are derived from experiments conducted using the enablers at venues of the facility
EXPERIMEDIA: Innovate in New Media - a multi-venue experimentation service supporting technology innovation through new forms of social interaction and user experience
No full textNew media applications and services are revolutionising social interaction and user experience in both society and in wide ranging industry sectors. The rapid emergence of pervasive human and environment sensing technologies, novel immersive presentation devices and high performance, globally connected network and cloud infrastructures is generating huge opportunities for application providers, service provider and content providers. These new applications are driving convergence across devices, clouds, networks and services, and the merging of industries, technology and society. Yet the developers of such systems face many challenges in understanding how to optimise their solutions (Quality of Service - QoS) to enhance user experience (Quality of Experience - QoE) and how their disruptive innovations can be introduced into the market with appropriate business models. In this report, we present the results of a new multi-disciplinary collaborative approach to product and service innovation that brings together users, technology and live events in a series of experiments conducted in real world settings. Through experimentation we have explored a broad range of technical, societal and economic challenges faced by technology providers each aiming to create and exploit new multimedia value chains in markets such as leisure and tourism, cultural and heritage, and sports science and training. The experiments highlight the features of multimedia systems and the future opportunities for companies, as the Internet continues to transition towards the increasingly connected world of Internet of Things and Big Data. We know that putting user values at the heart of design decisions and evaluation is the key to success, and that long term benefits to providers of technology, services and content must derive from enhanced user experience. Engaging users in real-world settings to co-design and assess how technology can be used is now more important than testing how technology will be operated. We have only scratched the surface of possibility in novel networked multimedia systems yet we believe that the individual and collective results in the report are significant as they are grounded in real-world evidence. A new way of conducting research and innovation has been created that maximises the potential for commercial exploitation and societal impact. We think this is extremely important and when adopted will lead to greater benefits for all
EXPERIMEDIA – a multi-venue experimentation service supporting technology innovation through new forms of social interaction and user experience
No full textNew media applications and services are revolutionising social interaction and user experience in both society and in wide ranging industry sectors. The rapid emergence of pervasive human and environment sensing technologies, novel immersive presentation devices and high performance, globally connected network and cloud infrastructures is generating huge opportunities for application providers, service provider and content providers.These new applications are driving convergence across devices, clouds, networks and services, and the merging of industries, technology and society. Yet the developers of such systems face many challenges in understanding how to optimise their solutions (Quality of Service – QoS) to enhance user experience (Quality of Experience – QoE) and how their disruptive innovations can be introduced into the market with appropriate business models
Online classification of visual tasks for industrial workflow monitoring
No full textModelling and classification of time series stemming from visual workflows is a very challenging problem due to the inherent complexity of the activity patterns involved and the difficulty in tracking moving targets. In this paper, we propose a framework for classification of visual tasks in industrial environments. We propose a novel method to automatically segment the input stream and to classify the resulting segments using prior knowledge and hidden Markov models (HMMs), combined through a genetic algorithm. We compare this method to an echo state network (ESN) approach, which is appropriate for general-purpose time-series classification. In addition, we explore the applicability of several fusion schemes for multicamera configuration in order to mitigate the problem of limited visibility and occlusions. The performance of the suggested approaches is evaluated on real-world visual behaviour scenarios
Goat-CNN: A Lightweight Convolutional Neural Network for Pose-Independent Body Condition Score Estimation in Goats
No full text<p>Here we introduce the dataset utilized in our published paper entitled "<a href="https://www.sciencedirect.com/science/article/pii/S2666154324002114">Goat-CNN: A Lightweight Convolutional Neural Network for Pose-Independent Body Condition Score Estimation in Goats</a>".</p>
<p>Contained within the "bcs" folder are all the videos collected for this study. Each video file is named with a format denoting its respective details. The first number signifies the sequence of collection, the second denotes the ear tag, and the final figure represents the body condition score (BCS) value.</p>
<p>For example: "1_158734_2.50" indicates the first sampling of an animal with the ear tag "158734" and a BCS value of "2.50".</p>
<p>Additionally, we provide two Python scripts in this repository. The first script, "Video2Frame.py", facilitates the splitting of videos into individual frames. The second script, "Frames2npy.py", converts these frames into two numpy-friendly files with the extension ".npy". These files contain both the images ("X_train_bcs300.npy") and their corresponding labels ("Y_train_bcs300.npy").</p>
<p>Furthermore, for the convenience of swift experimentation, we have included the desired .npy files within the repository.</p>
<p>To load these files into your Python environment, you can use the following code snippet:</p>
<div>
<div>th4figs = '/content/drive/MyDrive/compag_2023/'</div>
<br>
<div>path4images = "/content/drive/MyDrive/CodeRefarm/datasets/BCS/X_train_bcs300.npy"</div>
<div>Xtrain = np.load(path4images)</div>
<br>
<div>path4labels = "/content/drive/MyDrive/CodeRefarm/datasets/BCS/Y_train_bcs300.npy"</div>
<div>Ytrain = np.load(path4labels).astype(float)</div>
<br>
<div>print("X train : ", Xtrain.shape)</div>
<div>print("Y train : ", Ytrain.shape)</div>
<div>
<div>
<div>
<div>
<div>
<div>
<div> </div>
</div>
<div> </div>
</div>
</div>
</div>
</div>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<pre>X train : (5332, 300, 300, 3)
Y train : (5332,)<br>
</pre>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div><p>If you want to cite this work you can use the following text:</p>
<p>@article{TEMENOS2024101174,<br>title = {Goat-CNN: A lightweight convolutional neural network for pose-independent body condition score estimation in goats},<br>journal = {Journal of Agriculture and Food Research},<br>volume = {16},<br>pages = {101174},<br>year = {2024},<br>issn = {2666-1543},<br>doi = {https://doi.org/10.1016/j.jafr.2024.101174},<br>url = {https://www.sciencedirect.com/science/article/pii/S2666154324002114},<br>author = {Anastasios Temenos and Athanasios Voulodimos and Vera Korelidou and Athanasios Gelasakis and Dimitrios Kalogeras and Anastasios Doulamis and Nikolaos Doulamis},<br>keywords = {Body condition score, Artificial intelligence, Convolutional neural network, Precision livestock farming, Goat, Animal, Signal processing, Computer vision},<br>abstract = {Modern livestock farming systems face the challenge of meeting the growing demand for dairy and meat products while ensuring the well-being of animals. Body Condition Scoring serves as a vital process for assessing the body reserves in animals, impacting their health, welfare, and productivity. However, traditional body condition score (BCS) evaluation methods via observation and palpation of specific anatomical regions are labor-intensive and subjective, hindering their widespread adoption. To address this issue, Precision Livestock Farming (PLF) techniques, particularly those involving Internet of Things (IoT) devices and artificial intelligence (AI), have emerged as promising solutions. In this work, we explore the use of AI, specifically Convolutional Neural Networks (CNNs), to automate the assessment of BCS in goats utilizing imagery data. Our model was trained on 5000 images illustrating the dorsal view of the backside of goats achieving an overall accuracy of 97.94 % which was the highest compared to other popular deep learning architectures from literature (e.g. VGG16, ResNet34, ResNet50, DenseNet, GoogleNet). The proposed custom CNN model for goat-specific BCS estimation overcomes the limitations of manual sketching, providing automatic region identification for BCS assessment. Moreover, it is a lightweight model specifically designed for seamless integration with IoT devices, allowing for efficient on-board processing via cameras. The model's pose-independent nature and adaptability to environmental constraints make it a valuable tool for efficient and sustainable goat farming. This research advances the application of AI as a precision livestock farming tool, contributing to the reinforcement of the animal welfare and productivity, and supporting evidence-based decision-making processes to increase farms' resilience.}<br>}</p>
<p>Anastasios Temenos, Athanasios Voulodimos, Vera Korelidou, Athanasios Gelasakis, Dimitrios Kalogeras, Anastasios Doulamis, Nikolaos Doulamis,<br>Goat-CNN: A lightweight convolutional neural network for pose-independent body condition score estimation in goats,<br>Journal of Agriculture and Food Research,<br>Volume 16,<br>2024,<br>101174,<br>ISSN 2666-1543,<br>https://doi.org/10.1016/j.jafr.2024.101174.<br>(https://www.sciencedirect.com/science/article/pii/S2666154324002114)<br>Abstract: Modern livestock farming systems face the challenge of meeting the growing demand for dairy and meat products while ensuring the well-being of animals. Body Condition Scoring serves as a vital process for assessing the body reserves in animals, impacting their health, welfare, and productivity. However, traditional body condition score (BCS) evaluation methods via observation and palpation of specific anatomical regions are labor-intensive and subjective, hindering their widespread adoption. To address this issue, Precision Livestock Farming (PLF) techniques, particularly those involving Internet of Things (IoT) devices and artificial intelligence (AI), have emerged as promising solutions. In this work, we explore the use of AI, specifically Convolutional Neural Networks (CNNs), to automate the assessment of BCS in goats utilizing imagery data. Our model was trained on 5000 images illustrating the dorsal view of the backside of goats achieving an overall accuracy of 97.94 % which was the highest compared to other popular deep learning architectures from literature (e.g. VGG16, ResNet34, ResNet50, DenseNet, GoogleNet). The proposed custom CNN model for goat-specific BCS estimation overcomes the limitations of manual sketching, providing automatic region identification for BCS assessment. Moreover, it is a lightweight model specifically designed for seamless integration with IoT devices, allowing for efficient on-board processing via cameras. The model's pose-independent nature and adaptability to environmental constraints make it a valuable tool for efficient and sustainable goat farming. This research advances the application of AI as a precision livestock farming tool, contributing to the reinforcement of the animal welfare and productivity, and supporting evidence-based decision-making processes to increase farms' resilience.<br>Keywords: Body condition score; Artificial intelligence; Convolutional neural network; Precision livestock farming; Goat; Animal; Signal processing; Computer vision</p>
Méthodes d'apprentissage automatique multimodales pour l'analyse de modèles dans les villes intelligentes et les transports
No full textDans le contexte des environnements urbains modernes et densément peuplés, la gestion efficace des transports et la structure des Systèmes de Transport Intelligents (STI) sont primordiales. Le secteur des transports publics connaît actuellement une expansion et une transformation significatives dans le but d'améliorer l'accessibilité, d'accommoder des volumes de passagers plus importants sans compromettre la qualité des déplacements, et d'adopter des pratiques respectueuses de l'environnement et durables. Les avancées technologiques, notamment dans l'Intelligence Artificielle (IA), l'Analyse de Données Massives (BDA), et les Capteurs Avancés (CA), ont joué un rôle essentiel dans la réalisation de ces objectifs et ont contribué au développement, à l'amélioration et à l'expansion des Systèmes de Transport Intelligents. Cette thèse aborde deux défis critiques dans le domaine des villes intelligentes, se concentrant spécifiquement sur l'identification des modes de transport utilisés par les citoyens à un moment donné et sur l'estimation et la prédiction du flux de transport au sein de divers systèmes de transport. Dans le contexte du premier défi, deux approches distinctes ont été développées pour la Détection des Modes de Transport. Tout d'abord, une approche d'apprentissage approfondi pour l'identification de huit médias de transport est proposée, utilisant des données de capteurs multimodaux collectées à partir des smartphones des utilisateurs. Cette approche est basée sur un réseau Long Short-Term Memory (LSTM) et une optimisation bayésienne des paramètres du modèle. À travers une évaluation expérimentale approfondie, l'approche proposée démontre des taux de reconnaissance remarquablement élevés par rapport à diverses approches d'apprentissage automatique, y compris des méthodes de pointe. La thèse aborde également des problèmes liés à la corrélation des caractéristiques et à l'impact de la réduction de la dimensionnalité. La deuxième approche implique un modèle basé sur un transformateur pour la détection des modes de transport appelé TMD-BERT. Ce modèle traite l'ensemble de la séquence de données, comprend l'importance de chaque partie de la séquence d'entrée, et attribue des poids en conséquence en utilisant des mécanismes d'attention pour saisir les dépendances globales dans la séquence. Les évaluations expérimentales mettent en évidence les performances exceptionnelles du modèle par rapport aux méthodes de pointe, soulignant sa haute précision de prédiction. Pour relever le défi de l'estimation du flux de transport, un Réseau Convolutif Temporel et Spatial (ST-GCN) est proposé. Ce réseau apprend à la fois des données spatiales du réseau de stations et des séries temporelles des changements de mobilité historiques pour prédire le flux de métro urbain et le partage de vélos à un moment futur. Le modèle combine des Réseaux Convolutifs Graphiques (GCN) et des Réseaux Long Short-Term Memory (LSTM) pour améliorer la précision de l'estimation. Des expériences approfondies menées sur des ensembles de données du monde réel du système de métro de Hangzhou et du système de partage de vélos de la ville de New York valident l'efficacité du modèle proposé, démontrant sa capacité à identifier des corrélations spatiales dynamiques entre les stations et à faire des prévisions précises à long terme.In the context of modern, densely populated urban environments, the effective management of transportation and the structure of Intelligent Transportation Systems (ITSs) are paramount. The public transportation sector is currently undergoing a significant expansion and transformation with the objective of enhancing accessibility, accommodating larger passenger volumes without compromising travel quality, and embracing environmentally conscious and sustainable practices. Technological advancements, particularly in Artificial Intelligence (AI), Big Data Analytics (BDA), and Advanced Sensors (AS), have played a pivotal role in achieving these goals and contributing to the development, enhancement, and expansion of Intelligent Transportation Systems. This thesis addresses two critical challenges within the realm of smart cities, specifically focusing on the identification of transportation modes utilized by citizens at any given moment and the estimation and prediction of transportation flow within diverse transportation systems. In the context of the first challenge, two distinct approaches have been developed for Transportation Mode Detection. Firstly, a deep learning approach for the identification of eight transportation media is proposed, utilizing multimodal sensor data collected from user smartphones. This approach is based on a Long Short-Term Memory (LSTM) network and Bayesian optimization of model’s parameters. Through extensive experimental evaluation, the proposed approach demonstrates remarkably high recognition rates compared to a variety of machine learning approaches, including state-of-the-art methods. The thesis also delves into issues related to feature correlation and the impact of dimensionality reduction. The second approach involves a transformer-based model for transportation mode detection named TMD-BERT. This model processes the entire sequence of data, comprehends the importance of each part of the input sequence, and assigns weights accordingly using attention mechanisms to grasp global dependencies in the sequence. Experimental evaluations showcase the model's exceptional performance compared to state-of-the-art methods, highlighting its high prediction accuracy. In addressing the challenge of transportation flow estimation, a Spatial-Temporal Graph Convolutional Recurrent Network is proposed. This network learns from both the spatial stations network data and time-series of historical mobility changes to predict urban metro and bike sharing flow at a future time. The model combines Graph Convolutional Networks (GCN) and Long Short-Term Memory (LSTM) Networks to enhance estimation accuracy. Extensive experiments conducted on real-world datasets from the Hangzhou metro system and the NY City bike sharing system validate the effectiveness of the proposed model, showcasing its ability to identify dynamic spatial correlations between stations and make accurate long-term forecasts
Virtual Reality Reconstruction Applications Standards for Maps, Artefacts, Archaeological Sites and Monuments
No full textVirtual Reality (abbreviated VR), although far from being a new concept in computer science, is increasingly considered nowadays as the digital media technology that can most directly linked with Archaeology and Archaeological Reconstruction (with the term “reconstruction” in this context being officially agreed on meaning the “re-building of a monument to its state at a time of its history chosen for that particular representation”). The potential along with the many degrees of freedom offered by this branch of technology in the sector recently led experts to even start talking about the dawn of the hyper-tourist era. The ever increasing amount of research in the area, as well as the number of actual archaeological sites that have been reconstructed in a VR environment to the present, appear to support both directly and indirectly such a strong statement. It is not an exaggeration to say that archaeological research is now dependent on VR more than ever before. Also, many of these applications include a pedagogical aspect in their design that makes them ideal educational platforms for students in archaeology and professionals in the area alike
Novel computer vision and machine learning techniques for behavior recognition from video and adaptation to a cloud computing environment
No full text205 σ.Αντικείμενο της διατριβής είναι η αναγνώριση ανθρώπινης συμπεριφοράς από ακολουθίες βίντεο με χρήση καινοτόμων τεχνικών όρασης υπολογιστών και μηχανικής μάθησης, με έμφαση σε δραστηριότητες οργανωμένες σε ροές εργασίας. Το πρόβλημα παρουσιάζει μια σειρά από σοβαρές δυσκολίες, όπως επικαλύψεις, συχνές αλλαγές φωτισμού και έκτοπα, ενώ τα οπτικώς σύνθετα περιβάλλοντα που εξετάζονται δημιουργούν πρόσθετες προκλήσεις. Οι τυπικές μέθοδοι που βασίζονται σε ανίχνευση ή ιχνηλάτηση αντικειμένων τείνουν να αποτυγχάνουν εξαιτίας της μεγάλης πολυπλοκότητας των αναπαριστώμενων σκηνών. Για να παρακαμφθούν αυτά τα στάδια, προτείνεται η εξαγωγή ολιστικών χαρακτηριστικών για την αναπαράσταση των εικόνων. Επίσης, εξετάζεται η δυνατότητα αξιοποίησης πληροφορίας από πολλαπλά ρεύματα παρατηρήσεων (κάμερες) μέσω τεχνικών σύμμιξης (fusion) των χρησιμοποιούμενων hidden Markov models (HMM) για την αντιμετώπιση των επικαλύψεων, ενώ ερευνάται η αποτελεσματικότητα της χρήσης της κατανομής Student-t αντί της Gaussian ως κατανομής παρατήρησης για μεγαλύτερη ευρωστία σε έκτοπα. Για το ερευνητικά και πρακτικά σημαντικό πρόβλημα της online αναγνώρισης συμπεριφοράς, προτείνεται μια σειρά από νέες τεχνικές με διαφορετική στόχευση. Η πρώτη στηρίζεται σε κατάτμηση των ακολουθιών, ταξινόμηση των προκυπτουσών υποακολουθιών με χρήση HMM και ενσωμάτωση a priori γνώσης μέσω γενετικού αλγορίθμου (GA-HMM). Η δεύτερη βασίζεται σε ένα συνδυασμό μπεϋζιανού φίλτρου και HMM και δεν απαιτεί ξεχωριστό αλγόριθμο κατάτμησης, ενώ η τρίτη αντιμετωπίζει το πρόβλημα ταυτόχρονων ή χρονικά επικαλυπτόμενων δραστηριοτήτων μέσω ανίχνευσης κίνησης σε περιοχές ενδιαφέροντος. Μια άλλη συνεισφορά της διατριβής έγκειται στην εισαγωγή τής έννοιας της Αξιολογικής Διόρθωσης, η οποία, αξιοποιώντας την ανατροφοδότηση που δίνεται από έναν εξειδικευμένο χρήστη ως προς την ορθότητα των προβλέψεων των αλγορίθμων αναγνώρισης, ακολουθεί μια προσέγγιση βασισμένη σε feedforward αλλά και προσαρμοστικά νευρωνικά δίκτυα με σκοπό τη μείωση του συνολικού σφάλματος ταξινόμησης και τη βελτίωση των μελλοντικών αποτελεσμάτων. Τέλος, εξετάζονται οι δυνατότητες αξιοποίησης των πλεονεκτημάτων του υπολογιστικού νέφους και προτείνεται μια αρχιτεκτονική βασισμένη σε πλατφόρμα νέφους με σκοπό την αποδοτικότερη και αποτελεσματικότερη εφαρμογή των προτεινόμενων μεθόδων σε πραγματικές εγκαταστάσεις υψηλής κλίμακας με απαιτήσεις online λειτουργίας σε πραγματικό χρόνο.This thesis aims at proposing novel computer vision and machine learning techniques for human behavior recognition from video, emphasizing on activities forming workflows. The problem involves significant challenges, such as occlusions, frequent illumination changes and outliers, whereas the visually complex environment of our use case induces additional difficulties. Typical object based methods that rely on detection or tracking tend to fail because of the high complexity of the observed scenes. In order to bypass these error-prone stages, we propose the extraction of holistic features directly on the image level for scene representation. Furthermore, we examine the applicability of fused schemes of the employed classifiers (hidden Markov models) in an endeavor to exploit redundancies from multiple camera streams so as to solve occlusions, whereas we also scrutinize the effectiveness of the multivariate Student-t distribution as observation likelihood (instead of the Gaussian) for higher tolerance to outliers. In the sequel, a series of new techniques for online behavior recognition are proposed. The first one is based on sequence segmentation, classification of the segments through HMM classifiers and incorporation of a priori knowledge via a genetic algorithm (GA-HMM). The second method is based on a combination of bayesian filtering and HMM and does not require a separate segmentation algorithm, while the third one focuses on concurrent activity recognition following a top-down event-driven Region of Interest based approach. Another contribution of this thesis lies in the introduction of the concept of Evaluative Rectification, which exploits an expert user’s feedback regarding the correctness of the classification results of the employed recognition methods and follows a feedforward and adaptive neural network based approach in order to improve future results in the direction of decreasing the overall classification error. Finally, we examine the possibility of exploiting the benefits involved in cloud computing and propose a cloud platform endowed with modern workflow management mechanisms for an effective and efficient application of the proposed methods in real-world large-scale installations, where online and real-time requirements are posed.Αθανάσιος Σ. Βουλόδημο
FAST-MDL: Fast Adaptive Supervised Training of multi-layered deep learning models for consistent object tracking and classification
No full text- …
