• Journal of the Korea Convergence Society
• /
• v.11 no.12
• /
• pp.39-47
• /
• 2020

In this paper, we proposed a deep learning-based detection method using Saliency Attention to detect salient objects in images. The salient object detection separates the object where the human eye is focused from the background, and determines the highly relevant part of the image. It is usefully used in various fields such as object tracking, detection, and recognition. Existing deep learning-based methods are mostly Autoencoder structures, and many feature losses occur in encoders that compress and extract features and decoders that decompress and extend the extracted features. These losses cause the salient object area to be lost or detect the background as an object. In the proposed method, Saliency Attention is proposed to reduce the feature loss and suppress the background region in the Autoencoder structure. The influence of the feature values was determined using the ELU activation function, and Attention was performed on the feature values in the normalized negative and positive regions, respectively. Through this Attention method, the background area was suppressed and the projected object area was emphasized. Experimental results showed improved detection results compared to existing deep learning methods.

• Journal of the Korea Convergence Society
• /
• v.12 no.11
• /
• pp.45-51
• /
• 2021

With the development of deep learning, semantic segmentation methods are being studied in various fields. There is a problem that segmenation accuracy drops in fields that require accuracy such as medical image analysis. In this paper, we improved PSPNet, which is a deep learning based segmentation method to minimized the loss of features during semantic segmentation. Conventional deep learning based segmentation methods result in lower resolution and loss of object features during feature extraction and compression. Due to these losses, the edge and the internal information of the object are lost, and there is a problem that the accuracy at the time of object segmentation is lowered. To solve these problems, we improved PSPNet, which is a semantic segmentation model. The multi-scale attention proposed to the conventional PSPNet was added to prevent feature loss of objects. The feature purification process was performed by applying the attention method to the conventional PPM module. By suppressing unnecessary feature information, eadg and texture information was improved. The proposed method trained on the Cityscapes dataset and use the segmentation index MIoU for quantitative evaluation. As a result of the experiment, the segmentation accuracy was improved by about 1.5% compared to the conventional PSPNet.

• Journal of the Korean Vacuum Society
• /
• v.15 no.4
• /
• pp.395-403
• /
• 2006

It is known that $MgAl_2O_4$ has higher resistance to moisture than MgO, in humid ambient MgO is chemically unstable. It reacts very easily with moisture in the air. In this study, the characteristic of $MgAl_2O_4$ and $MgAl_2O_4/MgO$ layers as dielectric protection layers for AC- PDP (Plasma Display Panel) have been investigated and analysed in comparison for conventional MgO layers. MgO and $MgAl_2O_4$ films both with a thickness of $1000\AA$ and $MgAl_2O_4/MgO$ film with a thickness of $200/800\AA$ were grown on the Cu substrates using the electron beam evaporation. $1000\AA$ thick aluminium layers were deposited on the protective layers in order to avoid the charging effect of $Ga^+$ ion beam while the focused ion beam(FIB) is being used. We obtained sputtering yieds for the MgO, $MgAl_2O_4$ and $MgAl_2O_4/MgO$ films using the FIB system. $MgAl_2O_4/MgO$ protective layers have been found th show $24{\sim}30%$ lower sputtering yield values from 0.244 up to 0.357 than MgO layers with the values from 0.364 up to 0.449 for irradiated $Ga^+$ ion beam with energies ranged from 10 kV to 14 kV. And $MgAl_2O_4$ layers have been found to show lowest sputtering yield values from 0.88 up to 0.109. Secondary electron emission coefficient(g) using the ${\gamma}$- FIB. $MgAl_2O_4/MgO$ and MgO have been found to have similar g values from 0.09 up to 0.12 for indicated $Ne^+$ ion with energies ranged from 50 V to 200 V. Observed images for the surfaces of MgO and $MgAl_2O_4/MgO$ protective layers, after discharge degradation process for 72 hours by SEM and AFM. It is found that $MgAl_2O_4/MgO$ protective layer has superior hardness and degradation resistance properties to MgO protective layer.

• Journal of Life Science
• /
• v.16 no.4
• /
• pp.612-617
• /
• 2006

Phosphoinositide-specific phospholipase $C-{\gamma}\;1\; (PLC-{\gamma}\;1)$ has pivotal roles in cellular signaling by producing second messengers, inositol 1,4,5-trisphosphate $(IP_3)$ and diacylglycerol (DG). Tubulin is a main component of microtubules and mitotic spindle fibers, which are composed of ${\alpha}-$ and ${\beta}-tubulin$ heterodimers in all eukaryotic cells. In humans, six ${\beta}-tubulin$ isotypes have been identified which display a distinct pattern of tissue expression. Previously we found that $PLC-{\gamma}\;1$ and one of four ${\beta}-tubulin$ isotypes including ${\beta}1$, ${\beta}2$, ${\beta}3$ and ${\beta}6$, colocalized in COS-7 cells and cotranslocated to the plasma membrane to activate $PLC-{\gamma}\;1$ upon agonist stimulation. In the present study, we demonstrate that the remaining two, tubulin ${\beta}4$ and ${\beta}5$, also showed a potential to activate $PLC-{\gamma}\;1$. The phosphatidylinositol 4,5-bisphosphate $(PIP_2)$ hydrolyzing activity of $PLC-{\gamma}\;1$ was substantially increased in the presence of purified ${\beta}4$ and ${\beta}5$ tubulin in vitro, whereas the activity was not promoted by bovine serum albumin, suggesting that tubulin ${\beta}4$ and ${\beta}5$ also activate $PLC-{\gamma}\;1$. Taken together, our results suggest that all the ${\beta}-tubulin$ isotype activates $PLC-{\gamma}\;1$ activity to regulate cellular signaling.

• Journal of Korean Ophthalmic Optics Society
• /
• v.20 no.3
• /
• pp.319-324
• /
• 2015

Purpose: This study was designed to evaluate the changes of phoria and calculated AC/A ratio, and their recovery time points by watching 3D television (3D TV). Methods: 50 subjects (male 30, female 20) of 20s to 40s ages who can watch 3D, were measured phoria using a Howell phoria card at 3 m for distance and 40 cm for near. The phoria was evaluated before watching 3D TV and every 10 minutes from starting of watching 3D TV for 30 minutes, and every 5 minutes after finishing of watching 3D TV for 30 minutes again. Results: For the distance phoria during and after watching 3D TV, it was increased to more exophoria $-0.98{\pm}1.37{\Delta}$ (prism diopters) after 10 minutes from starting of 3D TV watching (p=0.063) and increased to more exophoria $-1.00{\pm}1.28{\Delta}$ after 30 minutes (p=0.024), and started to decrease after finishing of watching 3D TV and recovered to the level of before 3D TV watching ($-0.78{\pm}1.11{\Delta}$) after 20 minutes (p=0.32) with comparing to phoria of before watching 3D TV ($-0.80{\pm}1.12{\Delta}$). For the near phoria, it was also increased to more exophoria $-5.71{\pm}4.45{\Delta}$ after 10 minutes from starting of watching 3D TV (p=0.000) and $-6.58{\pm}4.36{\Delta}$ after 30 minutes (p=0.000), and started to decrease after finishing of watching 3D TV and recovered to the level of before watching 3D TV after 20 minutes ($-4.34{\pm}3.67{\Delta}$) (p=0.32) with comparing to the phoria of before watching 3D TV ($-4.36{\pm}3.66{\Delta}$). AC/A ratio was decreased from $4.92{\pm}1.17{\Delta}/D$ for before 3D TV watching to $4.11{\pm}1.50{\Delta}/D$ for after 30 minutes from starting of watching 3D TV (p=0.000), and increased after the end of watching 3D TV and recovered to the level of before 3D TV watching ($4.93{\pm}1.18{\Delta}/D$) after 25 minutes (p=0.598). Conclusions: During watching 3D TV at near, it showed a tendency of convergence insufficiency by decrease of calculated AC/A ratio as result that exophoria at near was higher increased than exophoria at distance. However, the increased exophoria at both near and distance was recovered to the level of base line after 25 minutes from the end of watching 3D TV. Through this study, it seems to need rational proposals of advice for watching 3D TV.