Sökning: "Visual Transformer"
Visar resultat 1 - 5 av 11 uppsatser innehållade orden Visual Transformer.
1. Where to Fuse
Master-uppsats, Lunds universitet/Matematisk statistikSammanfattning : 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
2. Evaluation of deep learning methods for industrial automation
Master-uppsats, Umeå universitet/Institutionen för datavetenskapSammanfattning : 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
3. Visual Bird's-Eye View Object Detection for Autonomous Driving
Master-uppsats, Linköpings universitet/DatorseendeSammanfattning : In the field of autonomous driving a common scenario is to apply deep learningmodels on camera feeds to provide information about the surroundings. A recenttrend is for such vision-based methods to be centralized, in that they fuse imagesfrom all cameras in one big model for a single comprehensive output. LÄS MER
4. Large-scale Exploratory Text Visualisation
Magister-uppsats, Linköpings universitet/Medie- och Informationsteknik; Linköpings universitet/Tekniska fakultetenSammanfattning : The amount of available text data has increased rapidly in the latest years, making it difficult for an everyday user to find relevant information. To solve this, NLP and visualisation methods have been developed for extracting valuable information from text and presenting it to the user. LÄS MER
5. Handwritten Text Recognition Using a Vision Transformer
Master-uppsats, Uppsala universitet/Institutionen för informationsteknologiSammanfattning : The aim of this project is to create a method for offline handwritten text recognition using a vision transformer. It consists of two parts, where the first one segments all words in a document into separate images and the second one which recognizes the word on each image. LÄS MER