Sökning: "large scale image generation"

Visar resultat 1 - 5 av 14 uppsatser innehållade orden large scale image generation.

  1. 1. Image Quality Assessment Pipeline and Semi-Automated Annotation method for Synthetic Data

    Master-uppsats, Göteborgs universitet/Institutionen för data- och informationsteknik

    Författare :Liam Le Tran; Edina Dedovic; [2023-10-24]
    Nyckelord :Facial Expression Recognition; FACS; Action Units; styleGAN2-ada; synthetic data; Image Quality Assessment; Multi-stage Pre-training; Pipeline Processing; Semi-automated Human Annotation;

    Sammanfattning : Predicting human emotions through facial expression, particularly in relation to medication field such as clinical trial settings, has garnered scientific interest in recent years due to significant understanding of the impact of treatment on emotions and social functioning. This thesis aims to improve performance of a FER model using large scale of synthetic data. LÄS MER

  3. 2. Procedural System for Urban Forest Generation : Image-based Natural Terrain Generation

    Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)

    Författare :Bosen Cheng; [2022]
    Nyckelord :datauppsättningar; neurala nätverk; blickdetektering; texttaggning;

    Sammanfattning : Procedural content generation is a common way to create model resources in the computer graphics area. It is beneficial for generating large-scale models such as buildings, forests, etc. This project focus on the generation of the urban forest, which is a case often overlooked. LÄS MER

  4. 3. Analyzing the Negative Log-Likelihood Loss in Generative Modeling

    Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)

    Författare :Aleix Espuña I Fontcuberta; [2022]
    Nyckelord :Generative modeling; Normalizing flows; Generative Adversarial Networks; MaximumLikelihood Estimation; Real Non-Volume Preserving flow; Fréchet Inception Distance; Misspecification; Generativa metoder; Normalizing flows; Generative adversarial networks; Maximum likelihood-metoden; Real non-volume preserving flow; Fréchet inception distance; felspecificerade modeller;

    Sammanfattning : Maximum-Likelihood Estimation (MLE) is a classic model-fitting method from probability theory. However, it has been argued repeatedly that MLE is inappropriate for synthesis applications, since its priorities are at odds with important principles of human perception, and that, e.g. LÄS MER

  5. 4. Performance Assessment of Networked Immersive Media in Mobile Health Applications with Emphasis on Latency

    Magister-uppsats, Blekinge Tekniska Högskola

    Författare :Emmanuel Adebayo; [2021]
    Nyckelord :Performance Assessment; Networked Immersive Media; Cloud Virtual Reality; Cloud Network; Edge Network.;

    Sammanfattning : Cloud VR/AR/MR (Virtual Reality, Augmented Reality, and Mixed Reality) services representa high-level architecture that combines large scale computer resources in a data-center structurestyle set up to render VR/AR/MR services using a combination of very high bandwidth, ultralow latency, high throughput, latest 5G (5th Generation) mobile networks to the end users.  VR refers to a three-dimensional computer-generated virtual environment made up ofcomputers, which can be explored by people for real time interaction. LÄS MER

  7. 5. Seeing Stars - Intensity Interferometry in the Laboratory and on the Ground

    Master-uppsats, Lunds universitet/Astronomi - Genomgår omorganisation

    Författare :Colin Carlile; [2021]
    Nyckelord :Stellar Physics Instrumentation Interferometry Cherenkov Telescope Array Intensity Interferometry; Physics and Astronomy;

    Sammanfattning : Context Since the time of Tycho Brahe astronomy has been an observing science defined by the quality of its instrumentation. The introduction of primitive telescopes by Galileo in 1609 began a period of development of ever-larger telescopes with greater measuring precision. LÄS MER