Tecniche di deep learning per il monitoraggio del processo nell’industria dello pneumatico

Il seguente elaborato affronta l'implementazione di un algoritmo che affronta un problema di controllo di processo in ambito industriale utilizzando algoritmi di object detection. Infatti, il progetto concordato con il professore Di Stefano si è svolto in collaborazione con l’azienda Pirelli, nell’ambito della produzione di pneumatici. Lo scopo dell'algoritmo implementato è di verificare il preciso orientamento di elementi grafici della copertura, utilizzati dalle case automobilistiche per equipaggiare correttamente le vetture. In particolare, si devono individuare delle scritte sul battistrada della copertura e identificarne la posizione rispetto ad altri elementi fissati su di essa. La tesi affronta questo task in due parti distinte: la prima consiste nel training di algoritmi di deep learning per il riconoscimento degli elementi grafici e del battistrada, la seconda è un decisore che opera a valle del primo sistema utilizzando gli output delle reti allenate.

Metodologie di Corrispondenza Stereo Basate su Deep Learning per Superfici Altamente Riflettenti e Trasparenti: Dataset e Architettura

Nell’ambito della Stereo Vision, settore della Computer Vision, partendo da coppie di immagini RGB, si cerca di ricostruire la profondità della scena. La maggior parte degli algoritmi utilizzati per questo compito ipotizzano che tutte le superfici presenti nella scena siano lambertiane. Quando sono presenti superfici non lambertiane (riflettenti o trasparenti), gli algoritmi stereo esistenti sbagliano la predizione della profondità. Per risolvere questo problema, durante l’esperienza di tirocinio, si è realizzato un dataset contenente oggetti trasparenti e riflettenti che sono la base per l’allenamento della rete. Agli oggetti presenti nelle scene sono associate annotazioni 3D usate per allenare la rete. Invece, nel seguente lavoro di tesi, utilizzando l’algoritmo RAFT-Stereo [1], rete allo stato dell’arte per la stereo vision, si analizza come la rete modifica le sue prestazioni (predizione della disparità) se al suo interno viene inserito un modulo per la segmentazione semantica degli oggetti. Si introduce questo layer aggiuntivo perché, trovare la corrispondenza tra due punti appartenenti a superfici lambertiane, risulta essere molto complesso per una normale rete. Si vuole utilizzare l’informazione semantica per riconoscere questi tipi di superfici e così migliorarne la disparità. È stata scelta questa architettura neurale in quanto, durante l’esperienza di tirocinio riguardante la creazione del dataset Booster [2], è risultata la migliore su questo dataset. L’obiettivo ultimo di questo lavoro è vedere se il riconoscimento di superfici non lambertiane, da parte del modulo semantico, influenza la predizione della disparità migliorandola. Nell’ambito della stereo vision, gli elementi riflettenti e trasparenti risultano estremamente complessi da analizzare, ma restano tuttora oggetto di studio dati gli svariati settori di applicazione come la guida autonoma e la robotica.

Dataset Fusion application on deep learning Object Detection task

Application of dataset fusion techniques to an object detection task, involving the use of deep learning as convolutional neural networks, to manage to create a single RCNN architecture able to inference with good performances on two distinct datasets with different domains.

Identificazione di caratteristiche rilevanti in capi di abbigliamento mediante deep-learning

Il mondo della moda è in continua e costante evoluzione, non solo dal punto di vista sociale, ma anche da quello tecnologico. Nel corso del presente elaborato si è studiata la possibilità di riconoscere e segmentare abiti presenti in una immagine utilizzando reti neurali profonde e approcci moderni. Sono state, quindi, analizzate reti quali FasterRCNN, MaskRCNN, YOLOv5, FashionPedia e Match-RCNN. In seguito si è approfondito l’addestramento delle reti neurali profonde in scenari di alta parallelizzazione e su macchine dotate di molteplici GPU al fine di ridurre i tempi di addestramento. Inoltre si è sperimentata la possibilità di creare una rete per prevedere se un determinato abito possa avere successo in futuro analizzando semplicemente dati passati e una immagine del vestito in questione. Necessaria per tali compiti è stata, inoltre, una approfondita analisi dei dataset esistenti nel mondo della moda e dei metodi per utilizzarli per l’addestramento. Il presente elaborato è stato svolto nell’ambito del progetto FA.RE.TRA. per il quale l'Università di Bologna svolge un compito di consulenza per lo studio di fattibilità su reti neurali in grado di svolgere i compiti menzionati.

A deep learning model for controlling the galvanizing process in a production line

In the industry of steelmaking, the process of galvanizing is a treatment which is applied to protect the steel from corrosion. The air knife effect (AKE) occurs when nozzles emit a steam of air on the surfaces of a steel strip to remove excess zinc from it. In our work we formalized the problem to control the AKE and we implemented, with the R&D dept.of MarcegagliaSPA, a DL model able to drive the AKE. We call it controller. It takes as input the tuple : a tuple of the physical conditions of the process line (t,h,s) with the target value of the zinc coating (c); and generates the expected tuple of (pres and dist) to drive the mechanical nozzles towards the (c). According to the requirements we designed the structure of the network. We collected and explored the data set of the historical data of the smart factory. Finally, we designed the loss function as sum of three components: the minimization between the coating addressed by the network and the target value we want to reach; and two weighted minimization components for both pressure and distance. In our solution we construct a second module, named coating net, to predict the coating of zinc resulting from the AKE when the conditions are applied to the prod. line. Its structure is made by a linear and a deep nonlinear “residual” component learned by empirical observations. The predictions made by the coating nets are used as ground truth in the loss function of the controller. By tuning the weights of the different components of the loss function, it is possible to train models with slightly different optimization purposes. In the tests we compared the regularization of different strategies with the standard one in condition of optimal estimation for both; the overall accuracy is ± 3 g/m^2 dal target for all of them. Lastly, we analyze how the controller modeled the current solutions with the new logic: the sub-optimal values of pres and dist can be optimize of 50% and 20%.

Advances in scaling deep learning algorithms

Les algorithmes d'apprentissage profond forment un nouvel ensemble de méthodes puissantes pour l'apprentissage automatique. L'idée est de combiner des couches de facteurs latents en hierarchies. Cela requiert souvent un coût computationel plus elevé et augmente aussi le nombre de paramètres du modèle. Ainsi, l'utilisation de ces méthodes sur des problèmes à plus grande échelle demande de réduire leur coût et aussi d'améliorer leur régularisation et leur optimization. Cette thèse adresse cette question sur ces trois perspectives. Nous étudions tout d'abord le problème de réduire le coût de certains algorithmes profonds. Nous proposons deux méthodes pour entrainer des machines de Boltzmann restreintes et des auto-encodeurs débruitants sur des distributions sparses à haute dimension. Ceci est important pour l'application de ces algorithmes pour le traitement de langues naturelles. Ces deux méthodes (Dauphin et al., 2011; Dauphin and Bengio, 2013) utilisent l'échantillonage par importance pour échantilloner l'objectif de ces modèles. Nous observons que cela réduit significativement le temps d'entrainement. L'accéleration atteint 2 ordres de magnitude sur plusieurs bancs d'essai. Deuxièmement, nous introduisont un puissant régularisateur pour les méthodes profondes. Les résultats expérimentaux démontrent qu'un bon régularisateur est crucial pour obtenir de bonnes performances avec des gros réseaux (Hinton et al., 2012). Dans Rifai et al. (2011), nous proposons un nouveau régularisateur qui combine l'apprentissage non-supervisé et la propagation de tangente (Simard et al., 1992). Cette méthode exploite des principes géometriques et permit au moment de la publication d'atteindre des résultats à l'état de l'art. Finalement, nous considérons le problème d'optimiser des surfaces non-convexes à haute dimensionalité comme celle des réseaux de neurones. Tradionellement, l'abondance de minimum locaux était considéré comme la principale difficulté dans ces problèmes. Dans Dauphin et al. (2014a) nous argumentons à partir de résultats en statistique physique, de la théorie des matrices aléatoires, de la théorie des réseaux de neurones et à partir de résultats expérimentaux qu'une difficulté plus profonde provient de la prolifération de points-selle. Dans ce papier nous proposons aussi une nouvelle méthode pour l'optimisation non-convexe.

Applied Graph Theory in Computer Vision and Pattern Recognition

This book will serve as a foundation for a variety of useful applications of graph theory to computer vision, pattern recognition, and related areas. It covers a representative set of novel graph-theoretic methods for complex computer vision and pattern recognition tasks. The first part of the book presents the application of graph theory to low-level processing of digital images such as a new method for partitioning a given image into a hierarchy of homogeneous areas using graph pyramids, or a study of the relationship between graph theory and digital topology. Part II presents graph-theoretic learning algorithms for high-level computer vision and pattern recognition applications, including a survey of graph based methodologies for pattern recognition and computer vision, a presentation of a series of computationally efficient algorithms for testing graph isomorphism and related graph matching tasks in pattern recognition and a new graph distance measure to be used for solving graph matching problems. Finally, Part III provides detailed descriptions of several applications of graph-based methods to real-world pattern recognition tasks. It includes a critical review of the main graph-based and structural methods for fingerprint classification, a new method to visualize time series of graphs, and potential applications in computer network monitoring and abnormal event detection.

Designing a Computer Vision System for Underwater Robotic Interventions

This thesis deals with the challenging problem of designing systems able to perceive objects in underwater environments. In the last few decades research activities in robotics have advanced the state of art regarding intervention capabilities of autonomous systems. State of art in fields such as localization and navigation, real time perception and cognition, safe action and manipulation capabilities, applied to ground environments (both indoor and outdoor) has now reached such a readiness level that it allows high level autonomous operations. On the opposite side, the underwater environment remains a very difficult one for autonomous robots. Water influences the mechanical and electrical design of systems, interferes with sensors by limiting their capabilities, heavily impacts on data transmissions, and generally requires systems with low power consumption in order to enable reasonable mission duration. Interest in underwater applications is driven by needs of exploring and intervening in environments in which human capabilities are very limited. Nowadays, most underwater field operations are carried out by manned or remotely operated vehicles, deployed for explorations and limited intervention missions. Manned vehicles, directly on-board controlled, expose human operators to risks related to the stay in field of the mission, within a hostile environment. Remotely Operated Vehicles (ROV) currently represent the most advanced technology for underwater intervention services available on the market. These vehicles can be remotely operated for long time but they need support from an oceanographic vessel with multiple teams of highly specialized pilots. Vehicles equipped with multiple state-of-art sensors and capable to autonomously plan missions have been deployed in the last ten years and exploited as observers for underwater fauna, seabed, ship wrecks, and so on. On the other hand, underwater operations like object recovery and equipment maintenance are still challenging tasks to be conducted without human supervision since they require object perception and localization with much higher accuracy and robustness, to a degree seldom available in Autonomous Underwater Vehicles (AUV). This thesis reports the study, from design to deployment and evaluation, of a general purpose and configurable platform dedicated to stereo-vision perception in underwater environments. Several aspects related to the peculiar environment characteristics have been taken into account during all stages of system design and evaluation: depth of operation and light conditions, together with water turbidity and external weather, heavily impact on perception capabilities. The vision platform proposed in this work is a modular system comprising off-the-shelf components for both the imaging sensors and the computational unit, linked by a high performance ethernet network bus. The adopted design philosophy aims at achieving high flexibility in terms of feasible perception applications, that should not be as limited as in case of a special-purpose and dedicated hardware. Flexibility is required by the variability of underwater environments, with water conditions ranging from clear to turbid, light backscattering varying with daylight and depth, strong color distortion, and other environmental factors. Furthermore, the proposed modular design ensures an easier maintenance and update of the system over time. Performance of the proposed system, in terms of perception capabilities, has been evaluated in several underwater contexts taking advantage of the opportunity offered by the MARIS national project. Design issues like energy power consumption, heat dissipation and network capabilities have been evaluated in different scenarios. Finally, real-world experiments, conducted in multiple and variable underwater contexts, including open sea waters, have led to the collection of several datasets that have been publicly released to the scientific community. The vision system has been integrated in a state of the art AUV equipped with a robotic arm and gripper, and has been exploited in the robot control loop to successfully perform underwater grasping operations.

The use of electronic voting systems to enhance deep learning.

The present paper introduces a technology-enhanced teaching method that promotes deep learning. Four stages that correspond to four different student cohorts were used for its development and to analyse its effectiveness. The effectiveness of the method has been assessed in terms of examination results as well as results obtained from class response system software statistics. The evidence gathered indicates that the method developed is very effective and its implementation is straightforward. Furthermore, its success in achieving results seems to be independent of the skills and/or experience of the lecturer.

Supporting Scientific Research Through Machine and Deep Learning: Fluorescence Microscopy and Operational Intelligence Use Cases

Although the debate of what data science is has a long history and has not reached a complete consensus yet, Data Science can be summarized as the process of learning from data. Guided by the above vision, this thesis presents two independent data science projects developed in the scope of multidisciplinary applied research. The first part analyzes fluorescence microscopy images typically produced in life science experiments, where the objective is to count how many marked neuronal cells are present in each image. Aiming to automate the task for supporting research in the area, we propose a neural network architecture tuned specifically for this use case, cell ResUnet (c-ResUnet), and discuss the impact of alternative training strategies in overcoming particular challenges of our data. The approach provides good results in terms of both detection and counting, showing performance comparable to the interpretation of human operators. As a meaningful addition, we release the pre-trained model and the Fluorescent Neuronal Cells dataset collecting pixel-level annotations of where neuronal cells are located. In this way, we hope to help future research in the area and foster innovative methodologies for tackling similar problems. The second part deals with the problem of distributed data management in the context of LHC experiments, with a focus on supporting ATLAS operations concerning data transfer failures. In particular, we analyze error messages produced by failed transfers and propose a Machine Learning pipeline that leverages the word2vec language model and K-means clustering. This provides groups of similar errors that are presented to human operators as suggestions of potential issues to investigate. The approach is demonstrated on one full day of data, showing promising ability in understanding the message content and providing meaningful groupings, in line with previously reported incidents by human operators.

Geometric Deep Learning per il denoising di mesh 3D

La crescente disponibilità di scanner 3D ha reso più semplice l’acquisizione di modelli 3D dall’ambiente. A causa delle inevitabili imperfezioni ed errori che possono avvenire durante la fase di scansione, i modelli acquisiti possono risultare a volte inutilizzabili ed affetti da rumore. Le tecniche di denoising hanno come obiettivo quello di rimuovere dalla superficie della mesh 3D scannerizzata i disturbi provocati dal rumore, ristabilendo le caratteristiche originali della superficie senza introdurre false informazioni. Per risolvere questo problema, un approccio innovativo è quello di utilizzare il Geometric Deep Learning per addestrare una Rete Neurale in maniera da renderla in grado di eseguire efficacemente il denoising di mesh. L’obiettivo di questa tesi è descrivere il Geometric Deep Learning nell’ambito del problema sotto esame.

A Methodology for GUI Layer Redefinition Through Virtualization and Computer Vision

Nowadays despite improvements in usability and intuitiveness users have to adapt to the proposed systems to satisfy their needs. For instance, they must learn how to achieve tasks, how to interact with the system, and fulfill system's specifications. This paper proposes an approach to improve this situation enabling graphical user interface redefinition through virtualization and computer vision with the aim of increasing the system's usability. To achieve this goal the approach is based on enriched task models, virtualization and picture-driven computing.

Searching patterns in painting images with computer vision techniques

This paper presents a pattern recognition method focused on paintings images. The purpose is construct a system able to recognize authors or art styles based on common elements of his work (here called patterns). The method is based on comparing images that contain the same or similar patterns. It uses different computer vision techniques, like SIFT and SURF, to describe the patterns in descriptors, K-Means to classify and simplify these descriptors, and RANSAC to determine and detect good results. The method are good to find patterns of known images but not so good if they are not.