• Animal Systematics, Evolution and Diversity
• /
• v.40 no.4
• /
• pp.266-275
• /
• 2024

Parrots, especially the Cacatua species, are a particular focus for trade because of their mimicry, plumage, and intelligence. Indeed, Cacatua species are imported most into Korea. To manage trade in wildlife, it is essential to identify the traded species. This is conventionally achieved by morphological identification by experts, but the increasing volume of trade is overwhelming them. Identification of parrots, particularly Cacatua species, is difficult due to their similar features, leading to frequent misidentification. There is thus a need for tools to assist experts in accurately identifying Cacatua species in situ. Deep learning-based object detection models, such as the You Only Look Once (YOLO) series, have been successfully employed to classify wildlife and can help experts by reducing their workloads. Among these models, YOLO versions 5 and 8 have been widely applied for wildlife classification. The later model normally performs better, but selecting and designing a suitable model remains crucial for custom datasets, such as wildlife. Here, YOLO versions 5 and 8 were employed to classify 13 Cacatua species in the image data. Images of these species were collected from eBird, iNaturalist, and Google. The dataset was divided, with 80% used for training and validation and 20% for evaluating model performance. Model performance was measured by mean average precision, with YOLOv5 achieving 0.889 and YOLOv8 achieving 0.919. YOLOv8 was thus better than YOLOv5 at detecting and classifying Cacatua species in the examined images. The model developed here could significantly support the management of the global trade in Cacatua species.

• Journal of the Semiconductor & Display Technology
• /
• v.23 no.3
• /
• pp.140-147
• /
• 2024

This paper explores the recent advancements in speech recognition technology, focusing on the integration of artificial intelligence to improve recognition accuracy in challenging environments, such as noisy or low-quality audio conditions. Traditional speech recognition methods often suffer from performance degradation in noisy settings. However, the application of deep neural networks (DNN) has led to significant improvements, enabling more robust and reliable recognition in various industries, including banking, automotive, healthcare, and manufacturing. A key area of advancement is the use of Silent Speech Interfaces (SSI), which allow communication through non-speech signals, such as visual cues or other auxiliary signals like ultrasound and electromyography, making them particularly useful for individuals with speech impairments. The paper further discusses the development of multi-modal speech recognition, combining both audio and visual inputs, which enhances recognition accuracy in noisy environments. Recent research into lip-reading technology and the use of deep learning architectures, such as CNN and RNN, has significantly improved speech recognition by extracting meaningful features from video signals, even in difficult lighting conditions. Additionally, the paper covers the use of self-supervised learning techniques, like AV-HuBERT, which leverage large-scale, unlabeled audiovisual datasets to improve performance. The future of speech recognition technology is likely to see further integration of AI-driven methods, making it more applicable across diverse industries and for individuals with communication challenges. The conclusion emphasizes the need for further research, especially in languages with complex morphological structures, such as Korean

• Korean Journal of Remote Sensing
• /
• v.34 no.3
• /
• pp.465-480
• /
• 2018

On 20 January 2017, the fresh snow cover which is more than 20 cm, accompaning with lightning occurred over Yeongdong coastal region for the first 3-hour of the heavy snowfall event. This study analyzed sounding observations in the heavy snow period which were including the measurements of wind profiler, radiometer and rawinsonde. The features examined from the vertical wind and temperature data at the two adjacent stations, Bukgangneung and Gangneung-Wonju National University, are summarized as follows: 1) The strong (30-40 kts) north-east winds were observed in the level from 2 to 6 km. The Strong atmospheric instability was found from 4 to 6 km, in which the lapse rate of temperature was about $-18^{\circ}C\;km^{-1}$. These features indicate that the deep convective cloud develops up to the height of 6 km in the heavy snowfall period, which is shown in the satellite infrared images. 2) The cooling was observed in the level below 1 km. At this time, the surface air temperature at Bukgangneung station decreased by $4^{\circ}C$. The narrow cooling zone estimated from AWS and buoy data was located in east-west direction. These are the features observed in the cold front of extratropical cyclone. The distributions of radar echo and lightning also show the same shape in east-west direction. Therefore, the results indicate that the Yeongdong thundersnow event was the combined precipitation system of deep convective cloud and cold frontal precipitation.

• Journal of Chest Surgery
• /
• v.45 no.3
• /
• pp.171-176
• /
• 2012

Background: Deep neck infections (DNI) can originate from infection in the potential spaces and fascial planes of the neck. DNI can be managed without surgery, but there are cases that need surgical treatment, especially in the case of mediastinal involvement. The aim of this study is to identify clinical features of DNI and analyze the predisposing factors for mediastinal extension. Materials and Methods: We reviewed medical records of 56 patients suffering from DNI who underwent cervical drainage only (CD group) and those who underwent cervical drainage combined with mediastinal drainage for descending necrotizing mediastinitis (MD group) from August 2003 to May 2009 and compared the clinical features of each group and the predisposing factors for mediastinal extension. Results: Forty-four out of the 56 patients underwent cervical drainage only (79%) and 12 patients needed both cervical and mediastinal drainage (21%). There were no differences between the two groups in gender (p=0.28), but the MD group was older than the CD group (CD group, $44.2{\pm}23.2$ years; MD group, $55.6{\pm}12.1$ years; p=0.03). The MD group had a higher rate of co-morbidity than the CD group (p=0.04). The CD group involved more than two spaces in 14 cases (32%) and retropharyngeal involvement in 12 cases (27%). The MD group involved more than two spaces in 11 cases (92%) and retropharyngeal involvement in 12 cases (100%). Organism identification took place in 28 cases (64%) of the CD group and 3 cases of (25%) the MD group (p=0.02). The mean hospital stay of the CD group was $21.5{\pm}15.9$ days and that of the MD group was $41.4{\pm}29.4$ days (p=0.04). Conclusion: The predisposing factors of mediastinal extension in DNI were older age, involvement of two or more spaces, especially including the retropharyngeal space, and more comorbidities. The MD group had a longer hospital stay, higher mortality, and more failure to identify causative organisms of causative organisms than the CD group.

• The Bulletin of The Korean Astronomical Society
• /
• v.40 no.1
• /
• pp.46.1-46.1
• /
• 2015

M85 is a merger remnant galaxy in the Virgo Cluster, showing complex merging features. Globular clusters in M85 are a good tracer of its merging history. To investigate globular cluster system of M85, we obtain deep and wide field images of M85 in ugi filters covering one square degree using CFHT/MegaCam. We discover about 1,000 globular cluster candidates in these images. The color distribution of the globular cluster candidates within r < 5' from M85 does not show a clear bimodality and blue globular cluster candidates are more than red ones. These features are different from those in massive early-type galaxies. The spatial distribution of the globular cluster candidates is elongated along the faint stellar light of M85. We also investigate the spatial distribution of sub-populations of the globular cluster candidates with different color and brightness and estimate their ages based on their color. We discuss these results in relation with the formation history of M85.

• Journal of Korea Multimedia Society
• /
• v.8 no.3
• /
• pp.322-327
• /
• 2005

In this paper we Propose a method that enables to recognize the laser stripe with 3dimensional information of body. Laser stripe has 3-dimensional information. We found out patterns of stripe have features of body. So we made database of it using Fourier Descriptor method and compared it with another stripe of body to recognize bodies. We could recognize standard style of body efficiently It is respected that deep research should be studied on the different style of bodies and then the other features of human will be recognized.

• Journal of the Korean housing association
• /
• v.17 no.1
• /
• pp.97-105
• /
• 2006

The purposes of this study are to find out the current status and features of earth housing, and to explore users' level of satisfaction and needs of improvement about earth housing. Primary findings are as follows: (1) The area of earth housing is generally 25-34 pyong and the construction cost of earth housing is usually 3,000,000-3,400,000 won per pyong. (2) The age of earth house users is generally forties, fifties, and sixties. And their occupation is usually retiree and farmer. The age of earth based pension users is generally twenties and thirties. (3) The construction method of earth housing is usually earth brick structure reinforced with wood structure and earth brick structure(adobe). (4) The finish of outer wall is generally earth brick laying and earth plaster. And the finish of inner wall is usually wall paper and earth plaster. Roof tile and asphalt shingle is frequently observed as roof finish. (5) Users' satisfaction about earth housing is investigated high level. Especially, the satisfaction degrees about faculty of humidity control, stink elimination, prevention from sick house syndrome, support for psychological stabilization and deep sleep are observed very highly. (6) Reduction of construction cost and prevention of crack is investigated as needs of improvement about earth housing.

• Annals of Clinical Neurophysiology
• /
• v.8 no.2
• /
• pp.107-124
• /
• 2006

The basal ganglia are a group of nuclei located in the deep portion of the brain. Along with the cerebellum, the basal ganglia have a major role in controlling human voluntary movements, and their dysfunction is apparently responsible for various involuntary movements. Although the exact mechanism of how the basal ganglia control movements has yet to be clarified, the model of focused selection (through the direct pathway) and tonic inhibition (via the indirect pathway) is proposed to be a principal functional model of the basal ganglia. Parkinson's disease (PD) is classically characterized by bradykinesia, rigidity and tremor-at-rest. All features seem to be associated with dopamine depletion resulting from the degeneration of the nigrostriatal pathway, which produces reduced activity of the direct pathway and a concurrent enhancement of excitatory output from STN. This change may result in increased tonic background inhibition and reduced focused selection via the direct pathway, causing difficulties in performing voluntary movements selectively. However, it has not been possible to define a single underlying pathophysiologic mechanism that explains all parkinsonian symptoms. Here the data that give separate understanding to each of the three classic features are discussed.

• Biomedical Science Letters
• /
• v.18 no.4
• /
• pp.413-419
• /
• 2012

The melanoma differentiation-associated gene-7 (MDA-7) protein, also known as interleukin-24 (IL-24), is a novel candidate of tumor suppressor that can induce apoptosis experimentally in a variety of human malignant cells. However, there have been few studies about its role in colorectal cancer. We performed immunohistochemical detection of MDA-7/IL-24 in 399 tissue samples from primary colorectal adenocarcinoma patients using a tissue microarray. Western blotting was then done to confirm the immunohistochemical observations. MDA-7/IL-24 immunoreactivity was observed in 116 (29.1%) of the 399 colorectal adenocarcinoma cases. Analysis of the MDA-7/IL-24 expression by Western blotting confirmed the immunohistochemical results. The tumors with a negative MDA-7/IL-24 expression more frequently showed poor differentiation (P=0004), lymph node metastasis (P=0.001), deep invasion (P=0.008) and high stage (P=0.001). A subset of colorectal adenocarcinoma revealed a decreased expression of MDA-7/IL-24, and this was associated with progressive pathologic features. These findings suggest that loss of MDA-7/IL-24 expression may play a role in tumor growth and progression of colorectal adenocarcinomas.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.18 no.12
• /
• pp.38-44
• /
• 2019

Various automation studies have been conducted to detect defective products based on product images. In the case of machine vision-based studies, size and color error are detected through a preprocessing process. A situation may arise in which the main features are removed during the preprocessing process, thereby decreasing the accuracy. In addition, complex systems are required to detect various kinds of defects. In this study, we designed and developed a system to detect errors by analyzing various conditions of defective products. We designed the deep learning algorithm to detect the defective features from the product images during the automation process using a convolution neural network (CNN) and verified the performance by applying the algorithm to the checker-switch failure detection system. It was confirmed that all seven error characteristics were detected accurately, and it is expected that it will show excellent performance when applied to automation systems for error detection.