arXiv subjects

Sung-Jin Cho

Publications and source records attributed to Sung-Jin Cho.

5 recordsLinked to original sources

arXiv2021

Rethinking Coarse-to-Fine Approach in Single Image Deblurring

Coarse-to-fine strategies have been extensively used for the architecture design of single image deblurring networks. Conventional methods typically stack sub-networks with multi-scale input images and gradually improve sharpness of images from the bottom sub-network to the top sub-network, yielding inevitably high computational costs. Toward a fast and accurate deblurring network design, we revisit the coarse-to-fine strategy and present a multi-input multi-output U-net (MIMO-UNet). The MIMO-UNet has three distinct features. First, the single encoder of the MIMO-UNet takes multi-scale input images to ease the difficulty of training. Second, the single decoder of the MIMO-UNet outputs multiple deblurred images with different scales to mimic multi-cascaded U-nets using a single U-shaped network. Last, asymmetric feature fusion is introduced to merge multi-scale features in an efficient manner. Extensive experiments on the GoPro and RealBlur datasets demonstrate that the proposed network outperforms the state-of-the-art methods in terms of both accuracy and computational complexity. Source code is available for research purposes at https://github.com/chosj95/MIMO-UNet.

cs.CV↗

arXiv2019

W-Net: Two-stage U-Net with misaligned data for raw-to-RGB mapping

Recent research on learning a mapping between raw Bayer images and RGB images has progressed with the development of deep convolutional neural networks. A challenging data set namely the Zurich Raw-to-RGB data set (ZRR) has been released in the AIM 2019 raw-to-RGB mapping challenge. In ZRR, input raw and target RGB images are captured by two different cameras and thus not perfectly aligned. Moreover, camera metadata such as white balance gains and color correction matrix are not provided, which makes the challenge more difficult. In this paper, we explore an effective network structure and a loss function to address these issues. We exploit a two-stage U-Net architecture and also introduce a loss function that is less variant to alignment and more sensitive to color differences. In addition, we show an ensemble of networks trained with different loss functions can bring a significant performance gain. We demonstrate the superiority of our method by achieving the highest score in terms of both the peak signal-to-noise ratio and the structural similarity and obtaining the second-best mean-opinion-score in the challenge.

eess.IV↗

arXiv2010

Spin Oscillations in Antiferromagnetic NiO Triggered by Circularly Polarized Light

Coherent spin oscillations were non-thermally induced by circularly polarized pulses in fully compensated antiferromagnetic NiO. This effect is attributed to an entirely new mechanism of the action, on the spins, of the effective magnetic field generated by an inverse Faraday effect. The novelty of this mechanism is that spin oscillations are driven by the time derivative of the effective magnetic field acting even on "pure" antiferromagnets with zero net magnetic moment in the ground state. The measured frequencies (1.07 THz and 140 GHz) of the spin oscillations correspond to the out-of-plane and in-plane modes of antiferromagnetic magnons.

cond-mat.str-el↗

arXiv2005

A Magnetic and Moessbauer Spectral Study of Core/Shell Structured Fe/Au Nanoparticles

Fe/Au nanoparticles have been chemically synthesized through a reverse micelle reaction and investigated by both conventional and synchrotron based x-ray diffraction and by magnetic and Moessbauer spectral studies. The powder x-ray diffraction patterns reveal both the presence of crystalline alpha-iron and gold and the absence of any crystalline iron oxides or other crystalline products. First-order reversal curves, along with the major hysteresis loops of the Fe/Au nanoparticles have been measured as a function of time in order to investigate the evolution of their magnetic properties. The iron-57 Moessbauer spectra of both uncoated iron nanoparticles and the Fe/Au nanoparticles have been measured at 78 and 295 K and indicate that two major iron containing components are present, namely the expected alpha-iron and the unexpected amorphous Fe1-xBx alloy; several poorly crystallized ordered iron(III) oxide components as well as paramagnetic iron(II) and iron(III) components are also observed. These results indicate that the Fe-core/Au-shell nanoparticles synthesized through reverse micelles are far more complex that had been believed.

cond-mat.mtrl-sci↗

arXiv2005

Growth Mechanisms and Oxidation-Resistance of Gold-Coated Iron Nanoparticles

We report the chemical synthesis of Fe-core/Au-shell nanoparticles by a reverse micelle method, and the investigation of their growth mechanisms and oxidation-resistant characteristics. The core-shell structure and the presence of the Fe & Au phases have been confirmed by transmission electron microscopy, energy dispersive spectroscopy, X-ray diffraction, Mossbauer spectroscopy, and inductively coupled plasma techniques. Additionally, atomic-resolution Z-contrast imaging and electron energy loss spectroscopy (EELS) in a scanning transmission electron microscope (STEM) have been used to study details of the growth processes. The Au-shell grows by nucleating on the Fe-core surface before coalescing. The magnetic moments of such nanoparticles, in the loose powder form, decrease over time due to oxidation. The less than ideal oxidation-resistance of the Au shell may have been caused by the rough Au surfaces. However, in the pressed pellet form, electrical transport measurements show that the particles are fairly stable, as the resistance of the pellet does not change appreciably over time.

cond-mat.mtrl-sci↗