Browsing by Degree "Graduado o Graduada en Ingeniería Informática por la Universidad Pública de Navarra (Programa Internacional)"

Now showing 1 - 14 of 14
  • Thumbnail Image
    PublicationOpen Access
    3 Dimensional N Body Simulator
    (2022) Paños Basterra, Juan; Palacián Subiela, Jesús Francisco; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Telekomunikazio eta Industria Ingeniarien Goi Mailako Eskola Teknikoa
    This thesis is based on the development of a simulation software for the N-body problem, attempting to provide a modern program to improve upon existing simulators. To achieve this, the N-body problem will be explained, along with other related concepts such as numerical integrators. The development of the simulator will also be discussed, studying the selected framework, design principles followed, program structure and indepth code analysis. Finally, the results of the project will be studied, analyzing aspects such as performance, resulting UI and several improvements will be proposed. The results of this project are publicly available in https://github.com/panosjuanis/3DN-body-simulator, where anybody can download the developed software to utilize it, read the documentation, or contribute to it in whichever way they like.
  • Thumbnail Image
    PublicationOpen Access
    Comparing tangible and mid-air interfaces for interacting with a 3D audio controler
    (2021) Merino Pinedo, Fernando; Marzo Pérez, Asier; Eguinoa Cabrito, Rubén; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    3D Audio is becoming a new standard in society for music listening and general audio. New applications emerge every day trying to build each time more immersive experiences: videogames, environments simulations, virtual reality… where audio is a crucial part to reach this immersion. The department of Acoustics developed a 24-speakers metal sphere called JAULAB where users can go in and experiment real 3D audio. But we wanted more. Nowadays applications highlight by its user-machine interaction and freedom of action. That’s why, in a collaboration between the Acoustics Lab and UpnaLab, we would like to give users the opportunity to play and interact with the sphere to offer a much more immersive experience that could be used in the future for a great variety of applications (environment simulations, virtual reality, music production, music exhibition…) For this purpose, an implementation of two different interactive interfaces has been developed to be integrated in the sphere. The project has been divided in four different parts: the tangible interface implementation, the mid-air interface implementation, the sphere integration and finally the user study. We will be seeing all this parts during this report.
  • Thumbnail Image
    PublicationOpen Access
    eChess: magnetically actuated tokens with PCB coils
    (2021) Petkov Donkov, Stefan; Josu Irisarri; Marzo Pérez, Asier; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    The project consists in the design, construction and programming of a device capable of electromagnetically controlling the pieces of a chess board. The device will consist of two parts, one consists of a matrix of magnetic field sensors in charged of detecting the movement made by the user playing with the board, the other part has the function of moving the pieces around the board by the means of a matrix of printed circuit board coils controlled by a microcontroller. With this device a user would be capable of playing chess, physically against a computer that would make its moves automatically on the board. It would be possible to switch the game board and program the computer to play any other board game against it.
  • Thumbnail Image
    PublicationOpen Access
    Generation of invernal-valued fuzzy partitions in order to optimise IVFARC algorithm
    (2022) Otazu Redín, Judit; Sanz Delgado, José Antonio; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    IVFARC is a classifier based on interval-valued fuzzy association rules. This classifier provides its knowledge to correctly predict the class of a given example from a model based on fuzzy association rules. They are fuzzy because they define the space of fuzzy subsets, which allow to obtain the information of the problem. The first part of the algorithm is to generate the intervalvalued rules, so is needed to obtain the interval-valued fuzzy partitions of the data. This task is performed by a genetic algorithm, however, it is computationally very expensive and therefore very slow. The goal is to eliminate this first part of the algorithm and replace it with other ideas that are not so computationally demanding. An idea is proposed to use clustering methods to try to see the trend of the data and to define the mentioned interval-valued fuzzy partitions. Keeping in mind that directly interval-valued fuzzy sets must be obtained, the first thing to do is to find the centroids/representatives of the data. In case there are n linguistic labels, should be sought n centroids. This latter will construct the fuzzy sets, then by repeating this process several times and unifying the executions, will articulate the interval-valued fuzzy sets. Along with the interval-valued concept, it helps to allow for the management of uncertainty in the data and gives more flexibility in the rules.
  • No Thumbnail Available
    PublicationEmbargo
    Integration of a artificial intelligence model for the detection of emotions with Unreal Engine.
    (2021) Azcona Aizcorbe, Mario; Galar Idoate, Mikel; Fernández de Vigo, Carlos; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    When programming an Artificial Intelligence model in many cases the preferred programming language is Python, whereas the language utilized in Unreal Engine is C++. The objective of this project is taking an AI model programmed in Python and already trained and using it in Unreal Engine. It would also be possible to integrate in Unreal any Python program. This would make easier to program certain functionalities that could be implemented in Python rather than C++. The project will be carried out as a part of the bigger project Emotional Films, whose objective is to modify a film in real time based on the emotions of the viewer, which will be detected using images captured using a webcam.
  • No Thumbnail Available
    PublicationEmbargo
    Multi-camera set-up for generating deep learning training data: study of hardware limitations and integration with the user intrface
    (2021) Pascual Casas, Rubén; Galar Idoate, Mikel; Paternain Dallo, Daniel; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    This project is part of the Emotional Films project from the Public University of Navarra. More specifically it encompasses the image/video collection phase for the training of deep learning models. The project will focus on a study of the hardware limitations in the generation of the database and more specifically on the design of part of the application and integration of the cameras with it. For this, the following topics will be discussed: Market study with the different webcam options to use in the project. Study of properties and configuration of webcams in Python. Analysis of USB bandwidth consumption by webcams. Carrying out synchronization tests for taking pictures and videos. Study of the time it takes to take pictures and save images with several cameras. Study of storage options and possible size of the database. Finally, the integration of multi-camera capture with the data capture application.
  • Thumbnail Image
    PublicationOpen Access
    Optimizing the fuzzy transform using key points for image compression
    (2021) Galafate De la llave, Mikel; Paternain Dallo, Daniel; Sesma Sara, Mikel; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    In this project we work with the F-Transform and its application to image compression. With images that contain a single object with a constant background as a target, the main idea is to try to make a better approximation of the inverse F-Transform finding a better partition for each image applying a prior key point detection so that the F-Transform can be focused on the objects in the image. We make an optimization of the code for both the F-Transform and the approximation of its inverse, propose a method to find a better fuzzy partition and optimize it based on the magnitude of the first derivative of the image and lastly propose different methods that separate the object from the image to try to improve the results.
  • No Thumbnail Available
    PublicationEmbargo
    A review of Active Learning methods for classification problems
    (2022) Bruned Alamán, Jorge; Galar Idoate, Mikel; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    Within the Artificial Intelligence field, and, more specifically, in the context of Machine Learning, the need for a set of labelled data is a common need to carry out the learning process of supervised models. However, the retrieval of such labelled data can be a rather arduous, expensive, and/or time-consuming task. In order to increase this process’ efficiency by reducing the number of required labelled data, the concept of Active Learning was introduced in the literature. The main idea behind it is that, given a set of unlabeled data, the most useful instances for the learning process are selected, therefore labelling only the most important examples, rather than the whole dataset or a random subset of it. This task is dealt with by means of different metrics, which allow us to quantify the representativeness and informativeness of each individual instance, with the objective of determining whether it should be labelled. In this project, we review several Active Learning methods for classification problems, implement the most relevant approaches and test them in a common experimental framework.
  • Thumbnail Image
    PublicationOpen Access
    SandDiffusion: image generation with depth conditioning for augmented reality SandBoxes
    (2024) Vons Vons, Igor Volodimir; Marzo Pérez, Asier; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    This TFG emerges from a collaboration grant signed with UPNA during the academic exercise of 2022-2023, and which main focus was diving into the new AI technologies that started surging at the beginning of set academic year. Its main objective is exploring a real case implementation into already existing technologies, allowing for a newer, richer, and more expressive experience, enhancing its usage without needing to resort to complex algorithms which may be both extremely difficult to produce and maintain, and whose results are not at the level of the ones achieved with AI.
  • Thumbnail Image
    PublicationOpen Access
    Simultaneous localization and Mapping with EKF and known correspondes
    (2022) Dervisheva Dailova, Neri; Pagola Barrio, Miguel; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    The Simultaneous Localization and Mapping (SLAM) algorithm is the basis for the most navigation systems in a large range of indoor, outdoor, air and underwater applications for both manned and autonomous vehicles. The purpose of this thesis is to investigate and build a model for the SLAM algorithm. Experiments are carried on the algorithm to analyze different values of the parameters and evaluate for which one of them the model gives the best results. The two main algorithms implemented in Python were the Extended Kalman Filter (EKF) algorithm with known correspondences and the Occupancy Grid Mapping algorithm. Several case studies with different parameter values for these algorithms were defined to conduct the research by performing simulations in the V-REP software. The obtained results prove that the built model meets the three important convergence properties of the EKF algorithm as a basis for solving the SLAM problem. Through a good estimation technique and making assumptions about the environment some systems can solve the problem quite well, but for now SLAM is still an open problem.
  • Thumbnail Image
    PublicationOpen Access
    Study of different KNN aldorithm versions
    (2022) Sospedra Legarda, Javier; Sanz Delgado, José Antonio; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    K-Nearest Neighbor algorithm has been proven to be a simple and effective method for classification problems in machine learning. This Final Degree Project is based on the investigation of the KNN (K-Nearest Neighbors) algorithm introducing some changes that improve said algorithm: fuzzy logic, fuzzy intervals and evolutionary algorithms. First, a model using fuzzy logic and a model with fuzzy intervals are created, which improve to a certain extent the accuracy of the original model (KNN). Next, the evolutionary algorithm is used to try to improve the performance of the model further. The problem is the time required for the convergence of this last algorithm. Therefore, it is intended to make a comparison between all these models and see the difference between them in both performance and time.
  • Thumbnail Image
    PublicationOpen Access
    Study on the application of different image preprocessing algorithms in image segmentation using deep learning techniques
    (2022) Velasco Rodríguez, Iñaki; Paternain Dallo, Daniel; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    Semantic segmentation is one of the most important fields of computer vision due to its applicability. In this case, semantic segmentation will be applied to images taken from vehicles knowing the importance of autonomous driving for our future. This project aims to achieve precise and fast results in the field of semantic segmentation with the complication of using less powerful and more affordable hardware than the one used nowadays. Deep learning techniques will be used to solve this problem. More concretely, a specific model of convolutional neural network will be trained and in charge of making the predictions, a U-Net. Different parameters of the U-Net will be changed to study how they affect the results. Furthermore, various image sizes, color spaces or reduction methods will be applied to study their impact on the speed and accuracy of the U-Net predictions. Finally, all those results will be compared in order to make a final decision in which is the best combination and which fields impact the most and how.
  • Thumbnail Image
    PublicationOpen Access
    Studying the perturbation-based oversampling technique for imbalanced classification problems
    (2023) Saalim, Mehsun Ihtiyan; Sanz Delgado, José Antonio; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    This study aims to implement, analyze, and compare the effectiveness of a novel technique known as Perturbation-Based Oversampling (POS). This technique is designed to address class imbalance in machine learning by augmenting the minority class instances by strategically perturbing features using a hyperparameter ’p’. Two additional variations, namely POS 1.0 and POS 2.0, have been proposed as extensions of the original POS approach. Detailed experiments have been conducted across diverse datasets, presenting a comprehensive performance evaluation in terms of precision when compared to a selection of established methods designed to tackle unbalanced classification challenges.
  • Thumbnail Image
    PublicationOpen Access
    The cost of lies
    (2021) Domínguez Jalle, Íñigo; Villadangos Alonso, Jesús; Escuela Técnica Superior de Ingeniería Industrial, Informática y de Telecomunicación; Industria, Informatika eta Telekomunikazio Ingeniaritzako Goi Mailako Eskola Teknikoa
    The aim of this Final Degree Project is to develop a video game technical demo to study, compare and implement features from fields such as Physics, Procedural Random Generation, Performance improvements and Artificial Intelligence among others, using an Object-oriented programming language. To accomplish this, the author has developed a playable 3D Videogame demo showing his skills on the field, algorithms studied, implemented and some additional comparations between techniques with their results and final conclusions. When it is built, the demo allows the player to play and experience a FirstPerson Shooter with a simple story and a totally procedural random level system where the user can progress and advance until completion. The project has been tested in different platforms and systems in order to analyze and implement performance improvements with the goal to reach 144 Frames per second, a smooth gameplay and overall good experience. The engine used in the project is Unity Engine 2020.3.8f1 and the language C#.