Results 31 to 40 of about 2,105 (150)
Self-Supervised Learning with Swin Transformers
CoRR, 2021 We are witnessing a modeling shift from CNN to Transformers in computer vision. In this work, we present a self-supervised learning approach called MoBY, with Vision Transformers as its backbone architecture. The approach basically has no new inventions, which is combined from MoCo v2 and BYOL and tuned to achieve reasonably high accuracy on ImageNet ...
Zhenda Xie +6 more
openaire +3 more sources
Classification of Solar Radio Spectrum Based on Swin Transformer
Universe, 2022 Solar radio observation is a method used to study the Sun. It is very important for space weather early warning and solar physics research to automatically classify solar radio spectrums in real time and judge whether there is a solar radio burst. As the
Jian Chen +5 more
doaj +1 more source
Swin–MRDB: Pan-Sharpening Model Based on the Swin Transformer and Multi-Scale CNN
Applied Sciences, 2023 Pan-sharpening aims to create high-resolution spectrum images by fusing low-resolution hyperspectral (HS) images with high-resolution panchromatic (PAN) images. Inspired by the Swin transformer used in image classification tasks, this research constructs
Zifan Rong +3 more
doaj +1 more source
Swin on Axes: Extending Swin Transformers to Quadtree Image Representations [PDF]
2024 IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW)In recent years, Transformer models have revolutionized machine learning. While this has resulted in impressive re-sults in the field of Natural Language Processing, Computer Vision quickly stumbled upon computation and memory problems due to the high resolution and dimensionality of the input data. This is particularly true for video, where the number
Marc Oliu +3 more
openaire +2 more sources
SwinIR: Image Restoration Using Swin Transformer [PDF]
2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), 2021 Image restoration is a long-standing low-level vision problem that aims to restore high-quality images from low-quality images (e.g., downscaled, noisy and compressed images). While state-of-the-art image restoration methods are based on convolutional neural networks, few attempts have been made with Transformers which show impressive performance on ...
Jingyun Liang +5 more
openaire +2 more sources
A Swin Transformer-based model for mosquito species identification
Scientific Reports, 2022 Mosquito transmit numbers of parasites and pathogens resulting in fatal diseases. Species identification is a prerequisite for effective mosquito control. Existing morphological and molecular classification methods have evitable disadvantages.
De-zhong Zhao +8 more
doaj +1 more source
Comparative analysis of transformer architectures for brain tumor classification [PDF]
Exploration of MedicineAim: Early and accurate diagnosis of brain tumors is critical for treatment success, but manual magnetic resonance imaging (MRI) interpretation has limitations.
Yigitcan Cakmak, Ishak Pacal
doaj +1 more source
Swin-APT: An Enhancing Swin-Transformer Adaptor for Intelligent Transportation
Applied Sciences, 2023 Artificial Intelligence has been widely applied in intelligent transportation systems. In this work, Swin-APT, a deep learning-based approach for semantic segmentation and object detection in intelligent transportation systems is presented. Swin-APT includes a lightweight network and a multiscale adapter network designed for image semantic segmentation
Yunzhuo Liu +4 more
openaire +2 more sources
GIScience & Remote Sensing, 2023 Automatic surface water body mapping using remote sensing technology is greatly meaningful for studying inland water dynamics at regional to global scales.
Donghui Ma +3 more
doaj +1 more source
Enhancing medical image segmentation with a multi-transformer U-Net [PDF]
PeerJVarious segmentation networks based on Swin Transformer have shown promise in medical segmentation tasks. Nonetheless, challenges such as lower accuracy and slower training convergence have persisted. To tackle these issues, we introduce a novel approach
Yongping Dan +3 more
doaj +2 more sources