|
|
Learning the Face Behind a Voice
Krušina, Josef ; Matějka, Pavel (referee) ; Plchot, Oldřich (advisor)
This work addresses the problem of mapping fixed representations (embeddings) of a speech signal to face embeddings and then generating a face from the mapped embedding using a generative adversarial network (GAN) that was trained for face generation. GANs are a type of neural networks that can generate data similar to the data they were trained on. The architecture of the proposed system is based on four components: a face embedding extractor, a voice embedding extractor, an algorithm on top of a GAN that can generate a face from a face embedding, and my mapping network used to map a voice embedding to a face embedding. The pre-trained neural networks FaceNet and SpeechBrain are adopted as embedding extractors. A model that uses a pre-trained StyleGAN2 is adopted for backward face generation. The contribution of this work is that it allows the extrapolation of a face from audio signal only.
|
|
|
Learning the Face Behind a Voice
Kyjonka, Mojmír ; Matějka, Pavel (referee) ; Plchot, Oldřich (advisor)
This thesis deals with face reconstruction based on voice. The state of the art of this problem is investigated and model for such problem is trained. Model used in this thesis is based on the work "Reconstructing faces from voices" which architecture is based on Generative Adversarial Network (GAN). In this work, we used VGGFace and VoxCeleb datasets, and additionally, we created a small audiovisual dataset of Czech speakers. This work was implemented using the Python scripting language and PyTorch library.
|
|
|
Learning to Generate Images with Convolutional Neural Networks
Kohút, Jan ; Kolář, Martin (referee) ; Hradiš, Michal (advisor)
The aim of this Bachelor's thesis is to design and analyze convolutional neural networks generating images of characters based on their parameters. Parameters of characters are type of char, font, colour of character, background colour, translation and rotation. Neural networks have created multidimensional representation of each parameter. Relations inside these representation are similar to relations inside parameters. Neural networks generate characters with new values of parameters based on interpolation between learned values of parameters. Neural networks are capable to generalize problem of generating images.
|
|
|
Unique Car Counting
Uhrín, Peter ; Špaňhel, Jakub (referee) ; Juránek, Roman (advisor)
Current systems for counting cars on parking lots usually use specialized equipment, such as barriers at the parking lot entrance. Usage of such equipment is not suitable for free or residential parking areas. However, even in these car parks, it can help keep track of their occupancy and other data. The system designed in this thesis uses the YOLOv4 model for visual detection of cars in photos. It then calculates an embedding vector for each vehicle, which is used to describe cars and compare whether the car has changed over time at the same parking spot. This information is stored in the database and used to calculate various statistical values like total cars count, average occupancy, or average stay time. These values can be retrieved using REST API or be viewed in the web application.
|
|
|
Concomitant Outgrowth Event
Prokop, Lukáš ; Cséfalvay,, András (referee) ; Mazanec, Martin (advisor)
The diploma project – a short animated film accompanied by an even briefer theoretical text – is in its whole a convoluted topology of themes and approaches supposedly enclosed in a single narrative guiding the viewer through its labyrinth. A mythologizing auto-fiction, thus, lies alongside attempts at a disintegration of the human-technic duality, the dislocation of purely ocular vision from the current notion of epistemology, or the realignment of our perception of GCI – all meticulously interpreted through a world-building experiment grounded in a strictly science-fictional (and theory-fueled) imagination.
|
|
|
Learning the Face Behind a Voice
Krušina, Josef ; Matějka, Pavel (referee) ; Plchot, Oldřich (advisor)
This work addresses the problem of mapping fixed representations (embeddings) of a speech signal to face embeddings and then generating a face from the mapped embedding using a generative adversarial network (GAN) that was trained for face generation. GANs are a type of neural networks that can generate data similar to the data they were trained on. The architecture of the proposed system is based on four components: a face embedding extractor, a voice embedding extractor, an algorithm on top of a GAN that can generate a face from a face embedding, and my mapping network used to map a voice embedding to a face embedding. The pre-trained neural networks FaceNet and SpeechBrain are adopted as embedding extractors. A model that uses a pre-trained StyleGAN2 is adopted for backward face generation. The contribution of this work is that it allows the extrapolation of a face from audio signal only.
|
|
|
Embedding the approach to war reporting
Folta, Adam ; Marjanovič, Teodor (advisor) ; Osvaldová, Barbora (referee)
In this bachelor thesis called Embedding the approach to war reporting I would like to bring coherent material about embedding. The first part of the text will cover the historical setting of embedding in context. It roots into the ancient times and through World War I. or Vietnam War the text terminates with Ukrainan crisis. Another part of the work will bring information about objectivity whether embedding shows complete and comlex reporting. If is it true that journalist, who spent a time with troops, will be affected by the troops, thus the objectivity will turn up a dilemma.. In the bachelors thesis I would like to present principle of journalist ethics, which are usefull for being in a conflict or crisis situation. The legislation is also part of the text and it is connected with rules, which have to be followed by embed. These rules are part of the bigger concept of ISAF organization. Another part of my bachelor thesis involves an attitude of Czech army and shows how Czech system works. In the practical part of the work I will publish the transcript of three interviews with journalists like František Šulc, Michal Kubal and Teodor Marjanovič. These journalists were embedded with Czech and American troops in the Second Gulf War in 2003, and during operations in Afghanistan. In my bachelor...
|
|
|
Universal metric spaces
Raška, Martin ; Hušek, Miroslav (advisor) ; Vejnar, Benjamin (referee)
The thesis covers the properties of isometric embeddings of metric spaces into the Urysohn universal space U (P.S. Urysohn, 1927) and its generalizations (M. Katětov, 1988). The examination of various metric properties of the space U leads to the question of extendability of the embedding ϕ: M → U from a subspace M of a space P onto an embedding Φ: P → U. We approach to this question in situation P = M ∪ {p} in finer form. If ϕ denotes an embedding M → U, let Rϕ denotes the set of images of the point p in U under all possible isometric extensions of the embedding ϕ (we call Rϕ the space of realizations). The main objective of this thesis is answering the following question: Which forms do the spaces Rϕ assume, if ϕ passes all embeddings of the space M into the space U? Corollary 1 and theorem 3 in the II. part of the thesis metrically characterize the family {Rϕ|ϕ: M → U}. We use previous results in part III in order to determine the number of classes of metrically equivalent embeddings of the space M into the space U. As a consequence, we obtain the result of J. Melleray about the homogeneity of the space U.
|
|
|
Embedding formative assesment in elementary school
Laubová, Kristýna ; Krčmářová, Tereza (advisor) ; Hejlová, Helena (referee)
This thesis deals with the topic of formative assessment embedding. The goal of the theoretical part is to map the school assessment in general, mention its functions, types, forms and language. The purpose of this work is also to describe the process of formative assesment embeding, which is the topic of this thesis, where at first the formative assesment will be characterised and then divided into several sub-parts, which characterise suitable strategies for its embedding. Next goal is to summarize the characterstics of a lower age pupil, where the focus will be on his/her cognitive, emotional and social developement. The theoretical part is concluded by a chapter dealing with the specifics of a beggining teacher. The goal of this chapter is to describe his/her features and skills, which he she should have. All the above mentioned topics will be defined on the base of the professional literature. The methodology used in this thesis will be defined in the empirical part. The aim of the empirical part is gradual embedding of the formative assessment elements, which will be led in the 4th year of primary school, where I have been currently working as a teacher for the first year. The active teacher research, where the elements of the formative assessment are being embedded, will run for the period...
|
|
|
Learning the Face Behind a Voice
Kyjonka, Mojmír ; Matějka, Pavel (referee) ; Plchot, Oldřich (advisor)
This thesis deals with face reconstruction based on voice. The state of the art of this problem is investigated and model for such problem is trained. Model used in this thesis is based on the work "Reconstructing faces from voices" which architecture is based on Generative Adversarial Network (GAN). In this work, we used VGGFace and VoxCeleb datasets, and additionally, we created a small audiovisual dataset of Czech speakers. This work was implemented using the Python scripting language and PyTorch library.
|
guest ::
