Training with Augmented Data: GAN-based Flame-Burning Image Synthesis for Fire Segmentation in Warehouse

2021 ◽  
Author(s):  
Zhikai Yang ◽  
Teng Wang ◽  
Leping Bu ◽  
Jineng Ouyang
Keyword(s):  
Image Synthesis

Image Synthesis by combining the Wavelets and Inverse Approaches

1999 ◽  
Vol 19 (Supplement1) ◽  
pp. 87-90
Author(s):  
D. SEKIJIMA ◽  
S. HAYANO ◽  
Y. SAITO ◽  
T.L. KUNII
Keyword(s):  
Image Synthesis

scholarly journals Resolved spectral variations of the centimetre-wavelength continuum from the ρ Oph W photo-dissociation-region

2021 ◽  
Author(s):  
Simon Casassus ◽  
Matías Vidal ◽  
Carla Arce-Tord ◽  
Clive Dickinson ◽  
Glenn J White ◽  
...  
Keyword(s):  
Molecular Cloud ◽  
Dust Emission ◽  
Image Synthesis ◽  
Microwave Emission ◽  
Radio Continuum ◽  
Continuum Emission ◽  
Parametric Image ◽  
Physical Conditions ◽  
Morphological Differences ◽  
Dust Grain

Abstract Cm-wavelength radio continuum emission in excess of free-free, synchrotron and Rayleigh-Jeans dust emission (excess microwave emission, EME), and often called ‘anomalous microwave emission’, is bright in molecular cloud regions exposed to UV radiation, i.e. in photo-dissociation regions (PDRs). The EME correlates with IR dust emission on degree angular scales. Resolved observations of well-studied PDRs are needed to compare the spectral variations of the cm-continuum with tracers of physical conditions and of the dust grain population. The EME is particularly bright in the regions of the ρ Ophiuchi molecular cloud (ρ Oph) that surround the earliest type star in the complex, HD 147889, where the peak signal stems from the filament known as the ρ Oph-W PDR. Here we report on ATCA observations of ρ Oph-W that resolve the width of the filament. We recover extended emission using a variant of non-parametric image synthesis performed in the sky plane. The multi-frequency 17 GHz to 39 GHz mosaics reveal spectral variations in the cm-wavelength continuum. At ∼30 arcsec resolutions, the 17-20 GHz intensities follow tightly the mid-IR, Icm∝I(8 μm), despite the breakdown of this correlation on larger scales. However, while the 33-39 GHz filament is parallel to IRAC 8 μm, it is offset by 15–20 arcsec towards the UV source. Such morphological differences in frequency reflect spectral variations, which we quantify spectroscopically as a sharp and steepening high-frequency cutoff, interpreted in terms of the spinning dust emission mechanism as a minimum grain size acutoff ∼ 6 ± 1 Å that increases deeper into the PDR.

scholarly journals Visual Thinking of Neural Networks: Interactive Text to Image Synthesis

IEEE Access ◽  
2021 ◽  
pp. 1-1
Author(s):  
Hyunhee Lee ◽  
Gyeongmin Kim ◽  
Yuna Hur ◽  
Heuiseok Lim
Keyword(s):  
Neural Networks ◽  
Image Synthesis ◽  
Visual Thinking ◽  
Interactive Text

SAM-GAN: Self-Attention supporting Multi-stage Generative Adversarial Networks for text-to-image synthesis

2021 ◽  
Vol 138 ◽  
pp. 57-67
Author(s):  
Dunlu Peng ◽  
Wuchen Yang ◽  
Cong Liu ◽  
Shuairui Lü
Keyword(s):  
Image Synthesis ◽  
Generative Adversarial Networks ◽  
Adversarial Networks ◽  
Multi Stage

scholarly journals Enhanced Magnetic Resonance Image Synthesis with Contrast-Aware Generative Adversarial Networks

2021 ◽  
Vol 7 (8) ◽  
pp. 133
Author(s):  
Jonas Denck ◽  
Jens Guehring ◽  
Andreas Maier ◽  
Eva Rothgang
Keyword(s):  
Magnetic Resonance ◽  
Image Synthesis ◽  
Generative Adversarial Networks ◽  
Data Generation ◽  
Mr Images ◽  
Generative Adversarial Network ◽  
Radiology Training ◽  
Repetition Time ◽  
Echo Time ◽  
Image Orientation

A magnetic resonance imaging (MRI) exam typically consists of the acquisition of multiple MR pulse sequences, which are required for a reliable diagnosis. With the rise of generative deep learning models, approaches for the synthesis of MR images are developed to either synthesize additional MR contrasts, generate synthetic data, or augment existing data for AI training. While current generative approaches allow only the synthesis of specific sets of MR contrasts, we developed a method to generate synthetic MR images with adjustable image contrast. Therefore, we trained a generative adversarial network (GAN) with a separate auxiliary classifier (AC) network to generate synthetic MR knee images conditioned on various acquisition parameters (repetition time, echo time, and image orientation). The AC determined the repetition time with a mean absolute error (MAE) of 239.6 ms, the echo time with an MAE of 1.6 ms, and the image orientation with an accuracy of 100%. Therefore, it can properly condition the generator network during training. Moreover, in a visual Turing test, two experts mislabeled 40.5% of real and synthetic MR images, demonstrating that the image quality of the generated synthetic and real MR images is comparable. This work can support radiologists and technologists during the parameterization of MR sequences by previewing the yielded MR contrast, can serve as a valuable tool for radiology training, and can be used for customized data generation to support AI training.

Multi-modal image synthesis combining content-style adaptive normalization and attentive normalization

2021 ◽  
Author(s):  
Heng Zhang ◽  
Yuanyuan Pu ◽  
Rencan Nie ◽  
Dan Xu ◽  
Zhengpeng Zhao ◽  
...  
Keyword(s):  
Image Synthesis

scholarly journals Image synthesis with adversarial networks: A comprehensive survey and case studies

2021 ◽  
Author(s):  
Pourya Shamsolmoali ◽  
Masoumeh Zareapoor ◽  
Eric Granger ◽  
Huiyu Zhou ◽  
Ruili Wang ◽  
...  
Keyword(s):  
Case Studies ◽  
Image Synthesis ◽  
Adversarial Networks ◽  
Comprehensive Survey
Sign in / Sign up

Export Citation Format

Share Document