Sökning: "vision transformer"

Visar resultat 1 - 5 av 41 uppsatser innehållade orden vision transformer.

  1. 1. Where to Fuse

    Master-uppsats, Lunds universitet/Matematisk statistik

    Författare :Lukas Petersson; [2024]
    Nyckelord :Technology and Engineering;

    Sammanfattning : This thesis investigates fusion techniques in multimodal transformer models, focusing on enhancing the capabilities of large language models in understanding not just text, but also other modalities like images, audio, and sensor data. The study compares late fusion (concatenating modality tokens after separate encoding) and early fusion (concatenating before encoding) techniques, examining their respective advantages and disadvantages. LÄS MER

  3. 2. Analyzing the Influence of Synthetic andAugmented Data on Segmentation Model

    Uppsats för yrkesexamina på avancerad nivå, Luleå tekniska universitet/Institutionen för system- och rymdteknik

    Författare :Alex Peschel; [2023]
    Nyckelord :Artificial Intelligence; Microorganisms; Segmentation; Synthesizing; Augmentation;

    Sammanfattning : The field of Artificial Intelligence (AI) has experienced unprecedented growth in recent years, thanks to the numerous applications related to speech recognition, natural language processing, and computer vision. However, one of the challenges facing AI is the requirement for large amounts of energy, time, and data to be effective and accurate. LÄS MER

  4. 3. Few-Shot Learning for Quality Inspection

    Uppsats för yrkesexamina på avancerad nivå, Högskolan i Halmstad/Akademin för informationsteknologi

    Författare :Jesper Palmér; Ahmad Alsalehy; [2023]
    Nyckelord :Few-Shot Learning; AI; Transformers; ViT Deviation; Vision Transformers;

    Sammanfattning : The goal of this project is to find a suitable Few-Shot Learning (FSL) model that can be used in a fault detection system for use in an industrial setting. A dataset of Printed Circuit Board (PCB) images has been created to train different FSL models. LÄS MER

  5. 4. Convolution-compacted visiontransformers forprediction of localwall heat flux atmultiple Prandtlnumbers in turbulentchannel flow

    Master-uppsats, KTH/Skolan för teknikvetenskap (SCI)

    Författare :Yuning Wang; [2023]
    Nyckelord :Turbulent flow; Heat transfer; Vision transformer; Convolutional neural network; Machine learning;

    Sammanfattning : Predicting wall heat flux accurately in wall-bounded turbulent flows is critical for a variety of engineering applications, including thermal management systems and energy-efficient designs. Traditional methods, which rely on expensive numerical simulations, are hampered by increasing complexity and extremly high computation cost. LÄS MER

  7. 5. Evaluation of deep learning methods for industrial automation

    Master-uppsats, Umeå universitet/Institutionen för datavetenskap

    Författare :Ragnar Onning; [2023]
    Nyckelord :artificial intelligence; machine learning; deep learning; cnn; transformer; swin; swin transformer;

    Sammanfattning : The rise and adaptation of the transformer architecture from natural language processing to visual tasks have proven a useful and powerful tool. Subsequent architectures such as visual transformers (ViT) and shifting window (SWIN) transformers have proven to be comparable and oftentimes exceed convolutional neural networks (CNNs) in terms of accuracy. LÄS MER