• The Journal of Korea Robotics Society
• /
• v.16 no.4
• /
• pp.336-344
• /
• 2021

This paper presents a robust deep learning-based human tracking framework in crowded environments. For practical human tracking applications, a target must be robustly tracked even in undetected or overcrowded situations. The proposed framework consists of two parts: robust deep learning-based human detection and tracking while recognizing the aforementioned situations. In the former part, target candidates are detected using Detectron2, which is one of the powerful deep learning tools, and their weights are computed and assigned. Subsequently, a candidate with the highest weight is extracted and is utilized to track the target human using a Kalman filter. If the bounding boxes of the extracted candidate and another candidate are overlapped, it is regarded as a crowded situation. In this situation, the center information of the extracted candidate is compensated using the state estimated prior to the crowded situation. When candidates are not detected from Detectron2, it means that the target is completely occluded and the next state of the target is estimated using the Kalman prediction step only. In two experiments, people wearing the same color clothes and having a similar height roam around the given place by overlapping one another. The average error of the proposed framework was measured and compared with one of the conventional approaches. In the error result, the proposed framework showed its robustness in the crowded environments.

• 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.

• Archives of Plastic Surgery
• /
• v.35 no.1
• /
• pp.36-41
• /
• 2008

Purpose: For facial reconstruction, skin color match is crucial to achieve great aesthetic result. Forehead flap and full thickness skin graft have been used for many years to reconstruct facial defect. Their results are aesthetically valuable with remarkable resemblance and harmony of the skin color between donor and recipient sites. The purpose of this study is to evaluate and compare the aesthetic outcome of the two methods as the analysis of skin color match. Methods: From January 1995 to December 2005, ten forehead flaps and ten full thickness skin grafts were performed. The reconstructed areas of forehead flaps were five noses and five eyelids. Recipient sites of full thickness skin grafts were seven eyelids, two noses and one forehead. In order to obtain the objective validity, the skin color of flap(or graft) and the recipient sites were measured by chromameter. The skin colors were quantified according to a three-dimensional coordinate system used in chromameter, L*(brightness), a*(redness), and b*(yellowness).Results: There was no significant color difference between forehead flap site and adjacent skin in all color values. On the other hand, the L* and b* values of graft sites were significantly lower than those of the adjacent skins. The a* values of graft sites were higher than those of the adjacent skins. Conclusion: This study reveals that skin color match of forehead flap is greater than that of full thickness skin graft. As forehead flap has adequate volume and great color match, it can be useful to reconstruct deep facial defect such as nasal defect. On the other hand, full thickness skin graft can be used for superficial defect like partial eyelid defect.

• The Research Journal of the Costume Culture
• /
• v.28 no.5
• /
• pp.669-676
• /
• 2020

The purpose of this research was to revive the bathochromic effects of wool fabrics by using natural dyestuffs with minimum heavy metallic mordants. The natural dyestuffs used in this research were the indigo plant, Phellodendron amurense, and Caesalpinia sappan. Sample no. 1 was pre-dyed five times with indigo. Sample no. 2 was pre-dyed five times with indigo and then once dyed with Phellodendron amurense. Following the same method as sample no. 2 with an additional Phellodendron, Sample no. 3 consisted of a pre-dye five times with indigo and twice with Phellodendron amurense. Sample no. 4 was pre-dyed six times with indigo and then once dyed with Caesalpinia sappan. Sample no. 5 followed the same method as no. 4 with an additional dye of Caesalpinia sappan. Sample no. 6 was pre-dyed five times with indigo and then once dyed with Phellodendron amurense and once dyed with Caesalpinia sappan. The results were as follows: first, all samples showed deeper colors. Second, according to the results of the surface K/S measurement, the surface K/S of wool fabrics was >20. Third, the results of lightfastness measurement showed superiority over grade 4 in samples no. 1, 2, 3, 5, and 6. However, sample no. 4 was grade 3. In the colorfastness to washing measurement, sample no. 2 showed greater superiority than grade 3-4, while samples no. 1 and 3 were grade 3. In addition, the colorfastness to dry cleaning for all samples was satisfactory or excellent by more than grade 3.

• Journal of the Korea Fashion and Costume Design Association
• /
• v.14 no.4
• /
• pp.63-72
• /
• 2012

In this study, Characteristics of colors on Chinese traditional costume in Tang, Song, Yuan, Ming, and Qing Dynasty were compared. And general colors of Chinese traditional costume and the viewpoint of them were considered. The purpose of this study were to apt colors of Chinese traditional costumes to the sensory appetite of modern people for new things, and make the recreation inspired by new senses be applied into the fashion industry effectively. The results were as followed : Firstly, colors distribution of 5 dynasties were different. In the Tang, R(red) was highest, YR(yellow red) and Y(yellow) followed. In the Song, YR was highest, Y and R followed and PB(purple blue) was increased, In the Yuan, R, YR, PB, B were concentrated at low rates, and other colors were increased, In the Ming, Y, R and B were concentrated at low rates orderly, and other colors were increased, In the Qing, PB was highest and Y, R, YR, B, P(purple) were increased. Secondly, tons distribution of 5 dynasties were different. In the Tang, L(light) was highest, P(pale) and Lgr(light grayish) followed. In the Song, P(pale) was highest, Lgr, Gr(grayish) and Dl(dull) followed In the Yuan, Dp(deep) was highest, P and Lgr followed, and other tones were increased, In the Ming, S(strong) was highest, DI and V(vivid) followed. Thirdly, The general colors of 5 dynasty were R, YR, Y, B, PB colors with vivid, deep and light tones. Forth, the viewpoint of color on Chinese traditional costume was based on the Yin-Yang and the Five Elements theory. Based on results of this study, the successive researches will be carried out about the unique colors of each nation and application traditional colors to modern fashion industry in order to the unique cultural sense.

• Publications of The Korean Astronomical Society
• /
• v.30 no.2
• /
• pp.409-411
• /
• 2015

Galaxy clusters, the largest gravitationally bound systems, are an important subject of study to place constraints on cosmological models. Moreover, they are excellent places to test galaxy evolution models in connection to their environments. To date, massive clusters have been found unexpectedly (Kang & Im 2009; Gonzales et al. 2012) and the evolution of galaxies in clusters is still controversial (Elbaz et al. 2007; Faloon et al. 2013). Finding galaxy cluster candidates at z > 1 in a wide, deep imaging survey data will enable us to solve such issues of modern extragalactic astronomy. We report new candidate galaxy clusters in one of the wide and deep survey fields, the European Large Area ISO Survey North1 (ELAIS-N1) and North2 (ELAIS-N2) fields, covering a sky area of $8.75deg^2$ and $4.85deg^2$ each. We also suggest a new useful color selection technique to separate z > 1 galaxies from low - z galaxies by combining multi-wavelength data.

• Publications of The Korean Astronomical Society
• /
• v.30 no.2
• /
• pp.405-407
• /
• 2015

We describe a survey of quasars in the early universe, beyond z ~ 5, which is one of the main science goals of the Infrared Medium-deep Survey (IMS) conducted by the Center for the Exploration of the Origin of the Universe (CEOU). We use multi-wavelength archival data from SDSS, CFHTLS, UKIDSS, WISE, and SWIRE, which provide deep images over wide areas suitable for searching for high redshift quasars. In addition, we carried out a J-band imaging survey at the United Kingdom InfraRed Telescope with a depth of ~23 AB mag and survey area of ${\sim}120deg^2$, which makes IMS a suitable survey for finding faint, high redshift quasars at z ~ 7. In addition, for the quasar candidates at z ~ 5.5, we are conducting observations with the Camera for QUasars in EArly uNiverse (CQUEAN) on the 2.1m telescope at McDonald Observatory, which has a custom-designed filter set installed to enhance the efficiency of selecting robust quasar candidate samples in this redshift range. We used various color-color diagrams suitable for the specific redshift ranges, which can reduce contaminating sources such as M/L/T dwarfs, low redshift galaxies, and instrumental defects. The high redshift quasars we are confirming can provide us with clues to the growth of supermassive black holes since z ~ 7. By expanding the quasar sample at 5 < z < 7, the final stage of the hydrogen reionization in the intergalactic medium (IGM) can also be fully understood. Moreover, we can make useful constraints on the quasar luminosity function to study the contribution of quasars to the IGM reionization.

• Science of Emotion and Sensibility
• /
• v.15 no.1
• /
• pp.57-64
• /
• 2012

The purpose of this study was to compare the visual sensibility of red shirts in university students of two countries, Korea and China in order to suggest the color planning for shirts. Seven tonal red shirts' visual sensibility and preferences were evaluated by the students of Korea and from China in the sense of sight. Evaluated data were analyzed by using frequency analysis, average, factor analysis and t-test. The visual sensibility of red shirts was classified into three factors: sporty, romantic and classic. Those factors of visual sensibility showed a significant difference according to nationality and gender. The Chinese students had a more sporty feel to the strong tone, and a more classic feel to the dark tone than the Korean students. The deep tone red shirts was reputed to be more romantic by the Korean students, but the light tone and the pale tone were considered more romantic by the Chinese students. Meanwhile, the effects of nationality and gender on seven tone of preferences for red shirts was significantly differed.

• Korean Journal of Plant Resources
• /
• v.31 no.4
• /
• pp.419-422
• /
• 2018

A Hosta cultivar 'Neulpureum 1' was bred at the Korea National Arboretum, which produces new cultivars using vegetative propagation techniques. The new cultivar 'Neulpureum 1' was derived by crossing Hosta minor with Hosta 'Krossa Regal'. Among the induced leaf-color- and shape-modified hosta plants, the plants that exhibited deep-green color and small-bended leaves were selected. 'Neulpureum 1' maintained the deep-green leaves for longer than Hosta 'Krossa Regal'. Additionally, the plant height of 'Neulpureum 1' was shorter than that of Hosta 'Krossa Regal' and several leaves were observed on the new cultivar; therefore, it is likely to be used as a pot plant. The botanical characteristics were investigated for three years beginning 2012. A Hosta 'Neulpureum 1' can prove to be useful as a material for a pot plant or as ground cover plant at half-shadow place.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.8
• /
• pp.301-310
• /
• 2021

The TATI model is a Traffic Accident Text to RGB Image model, which is a methodology proposed in this paper for predicting the severity of traffic accidents. Traffic fatalities are decreasing every year, but they are among the low in the OECD members. Many studies have been conducted to reduce the death rate of traffic accidents, and among them, studies have been steadily conducted to reduce the incidence and mortality rate by predicting the severity of traffic accidents. In this regard, research has recently been active to predict the severity of traffic accidents by utilizing statistical models and deep learning models. In this paper, traffic accident dataset is converted to color images to predict the severity of traffic accidents, and this is done via CNN models. For performance comparison, we experiment that train the same data and compare the prediction results with the proposed model and other models. Through 10 experiments, we compare the accuracy and error range of four deep learning models. Experimental results show that the accuracy of the proposed model was the highest at 0.85, and the second lowest error range at 0.03 was shown to confirm the superiority of the performance.