In recent years semantic segmentation models utilizing Convolutional Neural Networks (CNN) have seen significant success for multiple different segmentation problems. Models such as U-Net have produced promising results within the medical field for both regular 2D and volumetric imaging, rivalling some of the best classical segmentation methods. In this thesis we examined the possibility of using a convolutional neural network-based model to perform segmentation of discrete bone fragments in CT-volumes with segmentation-hints provided by a user. We additionally examined different classical segmentation methods used in a post-processing refinement stage and their effect on the segmentation quality. We compared the performance of our model to similar approaches and provided insight into how the interactive aspect of the model affected the quality of the result. We found that the combined approach of interactive segmentation and deep learning produced results on par with some of the best methods presented, provided there were adequate amount of annotated training data. We additionally found that the number of segmentation hints provided to the model by the user significantly affected the quality of the result, with convergence of the result around 8 provided hints.

In large scale productions of metal sheets, it is important to maintain an effective way to continuously inspect the products passing through the production line. The inspection mainly consists of detection of defects and tracking of ID numbers. This thesis investigates the possibilities to create an automatic inspection system by evaluating different machine learning algorithms for defect detection and optical character recognition (OCR) on metal sheet data. Digit recognition and defect detection are solved separately, where the former compares the object detection algorithm Faster R-CNN and the classical machine learning algorithm NCGF, and the latter is based on unsupervised learning using a convolutional autoencoder (CAE). The advantage of the feature extraction method is that it only needs a couple of samples to be able to classify new digits, which is desirable in this case due to the lack of training data. Faster R-CNN, on the other hand, needs much more training data to solve the same problem. NCGF does however fail to classify noisy images and images of metal sheets containing an alloy, while Faster R-CNN seems to be a more promising solution with a final mean average precision of 98.59%. The CAE approach for defect detection showed promising result. The algorithm learned how to only reconstruct images without defects, resulting in reconstruction errors whenever a defect appears. The errors are initially classified using a basic thresholding approach, resulting in a 98.9% accuracy. However, this classifier requires supervised learning, which is why the clustering algorithm Gaussian mixture model (GMM) is investigated as well. The result shows that it should be possible to use GMM, but that it requires a lot of GPU resources to use it in an end-to-end solution with a CAE.

Semantic segmentation is a key approach to comprehensive image data analysis. It can be applied to analyze 2D images, videos, and even point clouds that contain 3D data points. On the first two problems, CNNs have achieved remarkable progress, but on point cloud segmentation, the results are less satisfactory due to challenges such as limited memory resource and difficulties in 3D point annotation. One of the research studies carried out by the Computer Vision Lab at Linköping University was aiming to ease the semantic segmentation of 3D point cloud. The idea is that by first projecting 3D data points to 2D space and then focusing only on the analysis of 2D images, we can reduce the overall workload for the segmentation process as well as exploit the existing well-developed 2D semantic segmentation techniques. In order to improve the performance of CNNs for 2D semantic segmentation, the study has used input data derived from different modalities. However, how different modalities can be optimally fused is still an open question. Based on the above-mentioned study, this thesis aims to improve the multistream framework architecture. More concretely, we investigate how different singlestream architectures impact the multistream framework with a given fusion method, and how different fusion methods contribute to the overall performance of a given multistream framework. As a result, our proposed fusion architecture outperformed all the investigated traditional fusion methods. Along with the best singlestream candidate and few additional training techniques, our final proposed multistream framework obtained a relative gain of 7.3\% mIoU compared to the baseline on the semantic3D point cloud test set, increasing the ranking from 12th to 5th position on the benchmark leaderboard.

Este trabalho de mestrado propõe a aplicação de um controlador robusto recursivo, de um controlador robusto com critério H \'INFINITO\', obtido via Teoria dos Jogos e de um controlador H \'INFINITO\' obtido através de LMIs, em um modelo dinâmico de robô bípede com joelhos e tronco. Considera-se um bípede planar de caminhar passivo, ou seja, possuí um ciclo limite e sete graus de liberdade, estando sujeito a incertezas paramétricas e a distúrbios de torque externo. A principal vantagem do controlador robusto recursivo é a ausência de ajustes de parâmetros auxiliares, facilitando a sua implementação online. Já o controlador H \'INFINITO\' obtido através da Teoria dos Jogos possuí a vantagem de não precisar de cálculos offline, tais como solucionar Desigualdades Matriciais Lineares. Já o último controlador implementado possuí a vantagem de ter seu ganho variando no tempo, assim como o controlador robusto recursivo. Os resultados de simulação apresentados mostram a eficácia na implementação desses controladores em robôs bípedes, obtendo-se índices de desempenho que tendem a mostrar um melhor resultado do controlador H \'INFINITO\' obtido através da Teoria dos Jogos. / This work proposes the implementation of a recursive robust control, a robust control with H \'INFINITO\' criterion obtained from game theory and a robust control with H \'INFINITO\' criteria obtained from the solutions of LMIs, in a dynamic model of a biped robot with knees and torso. We consider a planar biped robot with passive walk, i.e. it has limit-cycle, and has seven degrees of freedom. Also it is subject to parametric uncertainties and external disturbances. The main advantage of the recursive robust controller is the absence of auxiliary parameter settings, facilitating its implementation online. Since the controller H \'INFINITO\' obtained from the game theory has the advantage of not needing an offline calculation, such as solving linear matrix inequalities. And the third one has the advantage of the variable gain in time, witch occur with the recursive control too. The simulation results show the effectiveness of these controllers and the performance index show a tendency of better results applying the H \'INFINITO\' obtained from game theory.

In this thesis, three different neural network architectures are investigated to detect the action of a shot within a football game using video data. The first architecture uses con- ventional convolution and pooling layers as feature extraction. It acts as a baseline and gives insight into the challenges faced during shot detection. The second architecture uses a pre-trained feature extractor. The last architecture uses three-dimensional convolution. All these networks are trained using short video clips extracted from football game video streams. Apart from investigating network architectures, different sampling methods are evaluated as well. This thesis shows that amongst the three evaluated methods, the ap- proach using MobileNetV2 as a feature extractor works best. However, when applying the networks to a video stream there are a multitude of challenges, such as false positives and incorrect annotations that inhibit the potential of detecting shots.

Este projeto apresenta o desenvolvimento de um controle longitudinal e lateral para um veículo terrestre de categoria pesada, usando o conceito de geração de curvas de Clothoids. O controle é em malha fechada, com realimentação de velocidade e posição global (X,Y) do veículo no plano bi-dimensional. Dentro de uma arquitetura de controle autônomo para um veículo, o controle longitudinal ajusta a velocidade de cruzeiro em função da trajetória e o lateral é responsável por regular a direção do volante e a sua correspondência para com os pneus, que por sua vez direcionam o veículo dentro da trajetória dada. Para este controle, para o modelo do veículo foi apenas considerado a estrutura do cavalo mecânico (conjunto monolítico formado pela cabine, motor e rodas de tração do caminhão), desprezando qualquer carga traseira engatado nele. Primeiramente será apresentada uma breve introdução abordando a história e projetos atuas de veículos autônomos, em seguida é feito uma revisão dos conceitos básicos usados no projeto. No capitulo seguinte é abordado o modelo matemático do veículo (cinemática e dinâmica) e logo em seguida teremos a secção que aborda sobre a estrutura de controle proposta. A seguir será apresentado a seção de discussão sobre a implementação e resultados práticos. Finalmente é apresentado a conclusão e uma breve descrição sobre trabalhos futuros. / This project presents the development of a longitudinal and lateral control for a Heavy Category Ground Vehicles, using the concept of generation of curves Clothoids. This control is closed loop with feedback speed and position (X,Y) ofvehicle in two-dimensional plane. Within an autonomous control architecture for a vehicle, the longitudinal control adjusts cruising speed on the path and the lateral control is responsible for regulating direction of steering wheel and its correspondence to the tires, which in turn drive the vehicle within the given path. For this control, the vehicle model we are only considering the horse (monolithic assembly formed by the cab, engine and truck drive wheels), disregarding any rear cargo engaged in it. First a brief introduction will be presented addressing the history and projects of autonomous vehicles, then it is made a review of the basic concepts used in the project. The next chapter is discussed the mathematical model of the vehicle (kinematics and dynamics) and soon we will have a section dealing on the proposed control structure.The following will show the discussion section on the implementation and practical results, then the conclusion and a brief description of future work.

Computational stereo estimates 3D structure by analyzing visual changes between two or more passive images of a scene that are captured from different viewpoints. It is a key enabler for ubiquitous autonomous systems, large-scale surveying, virtual reality, and improved techniques for compression, tracking, and object recognition. The fact that computational stereo is an under-constrained inverse problem causes many challenges. Its computational and memory requirements are high. Typical heuristics and assumptions, used to constrain solutions or reduce computation, prevent treatment of key realities such as reflection, translucency, ambient lighting changes, or moving objects in the scene. As a result, a general solution is lacking. Stochastic models are common in computational stereo, but stochastic algorithms are severely under-represented. In this dissertation I present two stochastic algorithms and demonstrate their advantages over deterministic approaches. I first present the Quality-Efficient Stochastic Sampling (QUESS) approach. QUESS reduces the number of match quality function evaluations needed to estimate dense stereo correspondences. This facilitates the use of complex quality metrics or metrics that take unique values at non-integer disparities. QUESS is shown to outperform two competing approaches, and to have more attractive memory and scaling properties than approaches based on exhaustive sampling. I then present a second novel approach based on the Hough transform and extend it with distributed ray tracing (DRT). DRT is a stochastic anti-aliasing technique common to computer rendering but which has not been used in computational stereo. I demonstrate that the DRT-enhanced approach outperforms the unenhanced approach, a competing variation that uses re-accumulation in the Hough domain, and another baseline approach. DRT’s advantages are particularly strong for reduced image resolution and/or reduced accumulator matrix resolution. In support of this second approach, I develop two novel variations of the Hough transform that use DRT, and demonstrate that they outperform competing variations on a traditional line segment detection problem. I generalize these two examples to draw broader conclusions, suggest future work, and call for a deeper exploration by the community. Both practical and academic gaps in the state of the art can be reduced by a renewed exploration of stochastic computational stereo techniques. / text

Στην παρούσα εργασία, γίνεται μια παραμετρική διερεύνηση για το σχεδιασμό αυτόνομων φωτοβολταϊκών συστημάτων για περιοχές της νοτίου Ελλάδος. Ο όρος «αυτόνομα», αναφέρεται στα φωτοβολταϊκά συστήματα τα οποία λαμβάνουν την ηλιακή ενέργεια, την μετατρέπουν σε ηλεκτρική ενέργεια και εν συνεχεία μέσα από διάφορες ηλεκτρικές διατάξεις (μετατροπείς DC/DC, DC/AC κ.ά.) τροφοδοτούν είτε απευθείας είτε μέσω συσσωρευτών, το απαιτούμενο φορτίο. Μερικές κύριοι παράμετροι που μελετώνται είναι η κλίση της φωτοβολταϊκής συστοιχίας, η επιφάνεια της, το μέγεθος συσσωρευτών κ.ά.. Η μελέτη αυτή λαμβάνει χώρα για εφτά πόλεις τις νοτίου Ελλάδας, με δεδομένα την ηλιακή ενέργεια που φθάνει σε κάθε πόλη για κάθε μήνα του χρόνου. Πιο συγκεκριμένα, στο κεφάλαιο 1, γίνεται μια εισαγωγή στην ηλεκτρική ενέργεια και στις ανανεώσιμες πηγές ενέργειας. Στο κεφάλαιο 2, μελετάται το φωτοβολταϊκό φαινόμενο και γενικότερα τα φωτοβολταϊκά κύτταρα. Στο κεφάλαιο 3, μελετάται η σύνδεση πολλών φωτοβολταϊκών κυττάρων για το σχηματισμό ενός πλαισίου, ενός πάνελ ή μιας συστοιχίας. Στο κεφάλαιο 4, γίνεται αναφορά στα διάφορα είδη συσσωρευτών που υπάρχουν στα φωτοβολταϊκά συστήματα. Στο επόμενο 5ο κεφάλαιο, μελετάται η ηλιακή ακτινοβολία και πως μεταβάλλεται αυτή ανάλογα με τη κλίση της επιφάνειας πρόσπτωσης, και επίσης ποια είναι η βέλτιστη κλίση της συστοιχίας. Τέλος, στα κεφάλαια 6 και 7, γίνεται η παραμετρική διερεύνηση για το σχεδιασμό αυτόνομων Φ/Β συστημάτων. Πιο συγκεκριμένα, στο 6ο μελετάται η επίδραση των διαφόρων παραμέτρων για ετήσια λειτουργία του συστήματος, ενώ στο 7ο, για θερινή λειτουργία. / The present work presents a research on the parameters of autonomous photovoltaic systems for regions of south Greece. The term "autonomous" is referred to photovoltaic systems which receive solar power, they convert it into electric power and after that, through various electrical devices (converters DC/DC, DC/AC etc.) they feed either directly, or through batteries, the required load. Some main parameters that are studied are the tilt angle and surface of the photovoltaic array, the size of the batteries, etc. This research is carried out for seven cities of south Greece, with incoming data the solar radiation that arrives in every city for every month of the year. 5 More specifically, in chapter 1, there is an introduction about electric power and renewable energy sources. In chapter 2, it is studied what the photovoltaic effect is and generally the photovoltaic cells. In chapter 3, there is a research about connecting many photovoltaic cells, which are more commonly known as panels or arrays. Chapter 4 is dealt with to the various types of batteries that are used in the photovoltaic systems. In chapter 5, the solar radiation is studied and how it changes depending on the tilt angle of the surface of incidence and also which is the optimal inclination of the array. Finally, in chapters 6 and 7, there is a parametric research on the design of autonomous PV systems. More precisely, in chapter 6 is studied the effect of various parameters on the design of the system when the system is operating during the whole year, while in chapter 7, when the system is operating only during the summer period.

Background: Video surveillance is a growing area where it can help with deterring crime, support investigation or to help gather statistics. These are just some areas where video surveillance can aid society. However, there is an improvement that could increase the efficiency of video surveillance by introducing tracking. More specifically, tracking between cameras in a network. Automating this process could reduce the need for humans to monitor and review since the tracking can track and inform the relevant people on its own. This has a wide array of usability areas, such as forensic investigation, crime alerting, or tracking down people who have disappeared. Objectives: What we want to investigate is the common setup of real-time multi-target multi-camera tracking (MTMCT) systems. Next up, we want to investigate how the components in an MTMCT system affect each other and the complete system. Lastly, we want to see how image enhancement can affect the MTMCT. Methods: To achieve our objectives, we have conducted a systematic literature review to gather information. Using the information, we implemented an MTMCT system where we evaluated the components to see how they interact in the complete system. Lastly, we implemented two image enhancement techniques to see how they affect the MTMCT. Results: As we have discovered, most often, MTMCT is constructed using a detection for discovering object, tracking to keep track of the objects in a single camera and a re-identification method to ensure that objects across cameras have the same ID. The different components have quite a considerable effect on each other where they can sabotage and improve each other. An example could be that the quality of the bounding boxes affect the data which re-identification can extract. We discovered that the image enhancement we used did not introduce any significant improvement. Conclusions: The most common structure for MTMCT are detection, tracking and re-identification. From our finding, we can see that all the component affect each other, but re-identification is the one that is mostly affected by the other components and the image enhancement. The two tested image enhancement techniques could not introduce enough improvement, but other image enhancement could be used to make the MTMCT perform better. The MTMCT system we constructed did not manage to reach real-time.

The concept of Computer Vision is not new or fresh. On contrary ideas have been shared and worked on for almost 60 years. Many use cases have been found throughout the years and various systems developed, but there is always a place for improvement. An observation was made, that methods used today are generally focused on a single purpose and implement expensive technology, which could be improved. In this report, we are going to go through an extensive research to find out if a professionally sold, expensive software, can be replaced by an off the shelf, low-cost solution entirely designed and developed in-house. To do that we are going to look at the history of Computer Vision, examples of applications, algorithms, and find general scenarios or computer vision problems which can be solved. We are then going take a step further and define solid use cases for each of the scenarios found. Finally, a prototype solution is going to be designed and presented. After analysing the results gathered we are going to reach out to the reader convincing him/her that such application can be developed and work efficiently in various areas saving investments to businesses.