Aprèn - Francisco Tarrés Ruiz

The aim of this project is to automate the calculation of the total time that the logos of the sponsoring brands of moto GP appear on the screen during the races. This document explains all the steps that have been followed to train an automatic object detection model for a specific database using RetinaNet. At the beginning, a brief explanation of the main concepts of deep learning is given and it is explained how convolutional neural networks and their kinds of layers, convolution and pooling, operate. Afterwards, it is presented a state of the art of the main classification and object detection systems, where RetinaNet has been chosen because, currently, it is one of the systems that provides better results. It should be noted that the main difference between classification and detection is that a detection system obtains the position of the object (rectangular region called bounding box) and indicates its typology and the classification system only indicates its typology. A database of images from 6 moto GP videos had to be created and labeled using Labelimg software. Labeling an image means drawing the bounding box and defining which brand the logo belongs to. The selected brands have been: Alpinestar, DHL, Repsol, GoPro, Michelin, RedBull, Monster, Tissot, Motul and BMW and a total of 16 classes have been created since one can have several forms of logos. Due to the fact that the database is not large enough to train a model from scratch, the weights of a pre-trained network have been used, this technique is known as transfer learnig. In addition, to avoid overfitting, the layers of one part of the architecture called the backbone have been frozen, which, in this case, has been used with Resnet50. Later, another model has been trained applying data augmentation, to improve the results obtained from the first model trained. Data augmentation is a technique that generates new examples by performing transformations on the images in the database. With this, it has obtained an accuracy of 83.3% and a mean Average Precision (mAP) of 65.33%. Finally, an application example called Brand Logo Monitoring in Moto GP has been developed, which, using the model trained with data augmentation, counts automatically the time of appearance of each brand in the MotoGP Grand Prix and give a final result with the total time.

Convolutional neural networks for classifying studio/non-studio frames in TV news programs

Master thesis

Nevot Ezpeleta, Alexis Dionisio

Tutor / Supervisor: Tarrés Ruiz, Francisco

Publication date: 2020

Castelldefels School of Telecommunications and Aerospace Engineering (EETAC)

View detail Access to the full text

This project has studied an unusual topic on image recognition and classification that can have numerous utilities on multimedia content. The approach studied on this project has been scenes recognition on TV news programs, which is a good starting point to distinguish scenes on visual content displayed on TV. The work in this project can be the first steps to segmentscenes on movies, series, documentaries, etc... The methodology used to build a program capable to recognise and classifyimages is Deep Learning, a subset of Machine Learning based on artificialneural networks that learn from representations of data obtained after performing some operations on the input images. As in Machine Learning, atraining process with multiple example outputs is necessary to make the system capable to figure out the rules to obtain those outputs (categories) from the given inputs (frames). To build an image recognition system, the most suitable framework (previous comparison has been made between some of them) has been installed to code the operations that lead to the image classification andthe results to evaluate its performance. Another important part to complete this project is the data collection and management, the data volume has to be large and diverse in order to increase the number and type of patterns or features learnt from the representations, and therefore, to increase the system's capability to generalize. It also has to be managed properly, in order to not difficult the learning process with additional, human errors. The last part of this project consists on training the classifier in the different architectures, check which input parameters lead to the best performance and compare the results to decide which of them is the most appropriate to classify TV news program images. After the training is finished, the system must be tested with never-seen data in order to check its real performance and obtain a proper evaluation.

Redes convolucionales espacio-temporales para la detección de signantes en vídeo

Bachelor thesis

Egea Cadiz, Mari Alba

Tutor / Supervisor: Tarrés Ruiz, Francisco

Publication date: 2020

Castelldefels School of Telecommunications and Aerospace Engineering (EETAC)

View detail Access to the full text

STCSC-OT: Examen: primavera 2019 Restricted access to the UPC academic community

Exam

García Lozano, Mario; Gilabert Pinal, Pere Lluís; Montoro López, Gabriel; Pérez Díaz de Cerio, David; Spadaro, Salvatore; Tarrés Ruiz, Francisco

Publication date: 2019

Castelldefels School of Telecommunications and Aerospace Engineering (EETAC)

SYSTEMS AND TECHNOLOGIES FOR COMMUNICATIONS IN SMART CITIES - 300301

Access to the full text

STCSC-OT: Examen: primavera 2019 Restricted access to the UPC academic community

Exam

García Lozano, Mario; Gilabert Pinal, Pere Lluís; Montoro López, Gabriel; Pérez Díaz de Cerio, David; Spadaro, Salvatore; Tarrés Ruiz, Francisco

Publication date: 2019

Castelldefels School of Telecommunications and Aerospace Engineering (EETAC)

SYSTEMS AND TECHNOLOGIES FOR COMMUNICATIONS IN SMART CITIES - 300301

Access to the full text

STCSC-OT: Control: primavera 2019 Restricted access to the UPC academic community

Exam

García Lozano, Mario; Gilabert Pinal, Pere Lluís; Montoro López, Gabriel; Pérez Díaz de Cerio, David; Spadaro, Salvatore; Tarrés Ruiz, Francisco

Publication date: 2019

Castelldefels School of Telecommunications and Aerospace Engineering (EETAC)

SYSTEMS AND TECHNOLOGIES FOR COMMUNICATIONS IN SMART CITIES - 300301

Access to the full text

STCSC-OT: Control: primavera 2019 Restricted access to the UPC academic community

Exam

García Lozano, Mario; Gilabert Pinal, Pere Lluís; Montoro López, Gabriel; Pérez Díaz de Cerio, David; Spadaro, Salvatore; Tarrés Ruiz, Francisco

Publication date: 2019

Castelldefels School of Telecommunications and Aerospace Engineering (EETAC)

SYSTEMS AND TECHNOLOGIES FOR COMMUNICATIONS IN SMART CITIES - 300301

Access to the full text

Examen final Restricted access to the UPC academic community

Exam

García Lozano, Mario; Gilabert Pinal, Pere Lluís; Montoro López, Gabriel; Pérez Díaz de Cerio, David; Rodríguez Gómez, Alejandro Antonio; Spadaro, Salvatore; Tarrés Ruiz, Francisco

Publication date: 2019

Castelldefels School of Telecommunications and Aerospace Engineering (EETAC)

SYSTEMS AND TECHNOLOGIES FOR COMMUNICATIONS IN SMART CITIES - 300301

Access to the full text