• Korean Journal of Remote Sensing
• /
• v.38 no.5_2
• /
• pp.685-694
• /
• 2022

Building information is essential for various urban spatial analyses. For this reason, continuous building monitoring is required, but it is a subject with many practical difficulties. To this end, research is being conducted to extract buildings from satellite images that can be continuously observed over a wide area. Recently, deep learning-based semantic segmentation techniques have been used. In this study, a part of the structure of the context-based ResU-Net was modified, and training was conducted to automatically extract a building from a 30 cm Worldview-3 RGB image using SpaceNet's building v2 free open data. As a result of the classification accuracy evaluation, the f1-score, which was higher than the classification accuracy of the 2nd SpaceNet competition winners. Therefore, if Worldview-3 satellite imagery can be continuously provided, it will be possible to use the building extraction results of this study to generate an automatic model of building around the world.

• Journal of Broadcast Engineering
• /
• v.27 no.6
• /
• pp.840-847
• /
• 2022

In a deep neural network (DNN)-based speech enhancement, using global and local input speech information is closely related to model performance. Recently, a nested U-Net structure that utilizes global and local input data information using multi-scale has bee n proposed. This nested U-Net was also applied to speech enhancement and showed outstanding performance. However, a single skip connection used in nested U-Nets must be modified for the nested structure. In this paper, we propose a multi-level skip connection (MLS) to optimize the performance of the nested U-Net-based speech enhancement algorithm. As a result, the proposed MLS showed excellent performance improvement in various objective evaluation metrics compared to the standard skip connection, which means th at the MLS can optimize the performance of the nested U-Net-based speech enhancement algorithm. In addition, the final proposed m odel showed superior performance compared to other DNN-based speech enhancement models.

• Smart Media Journal
• /
• v.13 no.4
• /
• pp.57-64
• /
• 2024

Attributes such as material, color, and fit in fashion images are important factors for consumers to purchase clothing. However, the process of classifying clothing attributes requires a large amount of manpower and is inconsistent because it relies on the subjective judgment of human operators. To alleviate this problem, there is a need for research that utilizes artificial intelligence to classify clothing attributes in fashion images. Previous studies have mainly focused on classifying clothing attributes for either tops or bottoms, so there is a limitation that the attributes of both tops and bottoms cannot be identified simultaneously in the case of full-body fashion images. In this study, we propose a deep learning model that can distinguish between tops and bottoms in fashion images and classify the category of each item and the attributes of the clothing material. The deep learning models ResNet and EfficientNet were used in this study, and the dataset used for training was 1,002,718 fashion images and 125 labels including clothing categories and material properties. Based on the weighted F1-Score, ResNet is 0.800 and EfficientNet is 0.781, with ResNet showing better performance.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.31 no.1
• /
• pp.45-53
• /
• 1995

Working depth of the model net was determined by using of the same experimental tank and the same model net that used in the forwarded report in a series studies. The depth of the net which indicates the depth of the head rope from the water surface, was determined by the photographs taken in front of the net mouth with the combination of towing velocity, warp length and distance between paired boats. The results obtained can be summarized as follows: 1. Working depth of model nets A and B was varied in the range of 0.09~1.66$m$,and 0.04~1.34$m$(which can be converted into 2.7~40.2$m$and 1.2~49.8$m$in the full-scale net) respectively, and the depth of model net A was slightly deeper than the depth of the model net B. 2. Working depth ($D$,which is appendixed m for the model net, f for the full-scale net, A and B for the types of the model nets) can be expressed as the function of towing velocity$V_t$, as in the model net($V_t$=$m$/$sec$) $D_{mA}$=(-1.99+0.65$L_w$) $e^{-1.72V_t}$ $D_{mA]$=(-1.91+1.04 $L_w$) $e^{2.88V_t}$ in the full-scale net($V_t$=$k$'$t$ $D_{fA}$=(-29.32+0.65$L_w$)$e^{0.40 V_t}$ $D_{fB}$=(-57.60+1.04$L_w$)$e^{-0.67 V_t}$ 3. Working depth 9$D$ appendixes are as same as the former) can be expressed as the function of warp length$L_w$) in the model net, and can be converted into full-scale net as in the model net ($V_t$=$m$/$sec$) $D_{mA}$=-0.99 $e^{-1.42V_t}$+0.67$e^{-1359V_t}$$L_w$ $D_{mB}$=-.258$e^{-3.77V_t}$+1.16$e^{-3.15V_t$ $L^w$, in the full-scale net($V_t$=k't) $D_{fA}$=-29.28$e^{-0.32V_t}$+0.67$e^{-0.37V_t$$L_w$ $D_{fB}$=-69.10$e^{-0.81V_t}$+1.16$e^{-0.72V_t}$$L_w$. 4. Working depth was gradually shallowed according to the increase of the distance between paired boats.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.26 no.3
• /
• pp.230-236
• /
• 1990

To examine the behavior of yellow tail Seriola quinqueradiata and of file fish Navodon modestus to set net, the 1/30 scale model net of the actual set net used in the east coast of Korea was made of netting and set in the indoor experimental tank(700cm W$\times$700cm L$\times$100cm D), of which the water depth was kept 70cm. Total length of yellow tail examined were 37~45cm and file fish were 18~21cm. The fishes were released as groups of 3 to 20 individuals. The number of fish entered into the pound net or the bag net, escaped from the net was counted every minute and accumulated for 10minutes to check the ratio entered and the ratio escaped. The result obtained can be summarized as follows: 1. In case of yellow tail, even though the ratio entered into the pound net was not so high, that into the bag net was comparatively high. Traces of the fish swimming were comparatively smooth and both entering into the pound net and escaping from it was easily done. 2. In case of file fish, even though the ratio entered into the pound net was higher than that of yellow tail, that into the bag net was lower than that into the pound net. Traces of the fish swimming were roundly meandered. Entering into the bag net was not so smoothly done, even though that into the playground was smoothly done. 3. Both yellow tail and file fish, the ratio escaped from the bag net was almost0 and that from the pound net was less than 10%.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.32 no.3
• /
• pp.217-222
• /
• 1996

The moving behaviour of anchovy schools(There were small schools of sardine and round-herring in this anchovy schools} in the bag - net of a large scale set - net was investigated using a scanning sonar, in the Kishihata set - net fishing ground located in Nanao city Ishigawa prefecture, Japan from Dec. 1st to Dec. 4th, 1992. The results obtained are summarized as follows ; 1. Moving behaviour of fish schools between two bag - nets was more in the daytime than in the nighttime. 2. The rate of entering fish schools from the first bag -- net to the second bag - net were 78% when the fish school in the second bag -- net existed. The result suggest that fish school existence in the second bag - net affected the rate of entering fish schools from the first bag - net to the second- bag - net. However, the time of stay of fish school that entered the second bag -- net was very short when the fish school in the second bag - net existed densely. The result suggest that a high fish density in the second bag - net might enlarge the escape rate 3. The majority of fish schools moved to the center and off shore side of the bag - net when fish schools entered the bag - net.

• Journal of Environmental Science International
• /
• v.12 no.7
• /
• pp.745-751
• /
• 2003

The three-dimensional eco-hydrodynamic model was applied to estimate the physical process in terms of nutrients and net uptake(or regeneration) rate of nutrients in Kamak Bay for scenario analysis to find proper management plan. The estimation results of the physical process in terms of nutrients shelved that transportation of nutrients is dominant in surface level while accumulation of nutrients is dominant in bottom level. In the case of dissolved inorganic nitrogen, the results showed that the net uptake rate was 0∼60 mg/㎡/day in surface level(0∼3m), and the net regeneration rate was 0.0∼10.0 mg/㎡/day in middle level(3∼6m) and above 10mg/㎡/day in bottom level(6m∼below). In the case of dissolved inorganic phosphorus, the net uptake rate was 0.0∼3.0 mg/㎡/day in surface level, and the net regeneration rate was 0.5∼1.5 mg/㎡/day in middle level and 1.0∼3.0 mg/㎡/day in bottom level. These results indicates that net uptake and transport of nutrients are occurred predominantly at the surface level and the net generation and accumulation are dominant at bottom level. Therefore, it is important to consider the re-supplement of nutrients due to regeneration of bottom water.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1992.04b
• /
• pp.436-443
• /
• 1992

Presented in the paper is a systematic approach to constructing a Petri net model of FAS (flexible assembly system). Petri net is widely used in modeling automated manufacturing systems. But, it found to be very difficult for an FA engineer to build a correct model of an FAS with Petri net symbols (ie, place, transition, and token) from the beginning. An automated manufacturing system in general is built from a set of "standard" hardware components. An FAS in particular is usually composed of assembly robots, work tables, conveyor lines, buffer storages, part feeders, etc. In the proposed modeling scheme, each type of standard resources is represented as a standard "module" which is a sub Petri net. Then, the model of a FAS can be conveniently constructed using the predefined modules the same way the FAS itself is built from the standard components. The network representation of a FAS is termed a JR-net (job resource relation net) which is easy to construct. This JR net is then mechanically converted to a formal Petri net (to simulate the behavior of the FAS). The proposed modeling scheme may easily be extended to the modeling of other types of automated manufacturing systems such as FMS and AS/RS.ch as FMS and AS/RS.

• Wind and Structures
• /
• v.30 no.3
• /
• pp.219-229
• /
• 2020

Net pressures on roofs and walls of buildings are dependent on the internal and external pressure fluctuations. The variation of internal and external pressures are influenced by the size and location of the openings. The correlation of external and internal pressure influences the net pressures acting on cladding on different parts of the roof and walls. The peak internal and peak external pressures do not occur simultaneously, therefore, a reduction can be applied to the peak internal and external pressures to obtain a peak net pressure for cladding design. A 1:200 scale wind tunnel model study was conducted to determine the correlations of external and internal pressures and effective reduction to net pressures (i.e., net pressure factors, F_C) for roof and wall cladding. The results show that external and internal pressures on the windward roof and wall edges are well correlated. The largest ${\mathcal{C}}_{{\check{p},net}$, highest correlation coefficient and the highest F_C are obtained for different wind directions within 90° ≤ θ ≤ 135°, where the large openings are on the windward wall. The study also gives net pressure factors F_C for areas on the roof and wall cladding for nominally sealed buildings and the buildings with a large windward wall opening. These factors indicate that a 5% to 10% reduction to the action combination factor, K_C specified in AS/NZS 1170.2(2011) is possible for some critical design scenarios.

• Journal of Advanced Navigation Technology
• /
• v.13 no.3
• /
• pp.351-357
• /
• 2009

In this paper, we studied an efficient inter-working method in which QualNet network simulator can import WAVE channel model and physical layer simulation module pre-designed by Matlab tool. At first, we investigated physical layer and communication medium simply designed in QualNet, then we suggested practical method for QualNet network simulator to adopt different type of physical layer simulation module in which detailed multi-path fading channel model and IEEE802.11p communication modem are designed. This work should be applied to linked simulation between upper layer and lower physical layer for total simulation from higher layer to lower physical layer related to next generation DSRC/WAVE specification.