• Journal of the Korean Institute of Landscape Architecture
• /
• v.50 no.6
• /
• pp.58-69
• /
• 2022

This study compares and analyses the impacts of holding domestic garden expos that will be fundamental for holding garden expos in the future. Satisfaction with the three sites of the 2013 Suncheon Bay International Gardening Expo, 2015 Seoul Gardening Expo, and 2022 Goyang International Flower Fair, as well as factors affecting the satisfaction in economic, social/cultural, environment/ecology, and operation aspects, were analyzed. As a result of the study, the satisfaction level of all three sites was high, with a value of 3.5 or higher. In particular, satisfaction with the 2013 Suncheon Bay International Gardening Expo was the highest. It was found that there was a difference between the satisfaction level of the 2013 Suncheon Bay International Gardening Expo and the 2015 Seoul Gardening Expo. As a regional festival, the 2013 Suncheon Bay International Garden Expo has acquired a high status due to the 'Suncheon Bay Garden' being designated as the first National Garden. It is thought that great satisfaction was obtained because economic, social, cultural, and environmental revitalization was achieved by matching the values of citizen participation and ecological conservation. As a result of the comparison of perception types affecting the satisfaction by garden fair, satisfaction at the 2013 Suncheon Bay International Gardening Expo and the 2015 Seoul Gardening Expo, it was found that both affected all four aspects. The 2022 Goyang International Flower Fair did not affect satisfaction in the operational aspect. This seems to be because the Goyang International Flower Fair is already a fixed local brand. As a result of analyzing the detailed factors of perception that affect satisfaction, the three target sites were commonly analyzed, including social and cultural factors, which attract cultural events, improve pride and affection for the region, and help educate children. In terms of environmental and ecological factors were analyzed as an inconvenience in life due to traffic congestion. It can be seen that it has the same meaning as the comparative analysis of the difference in factors on the satisfaction of the target site. There is no difference in the effect on satisfaction in terms of social·cultural, and environmental·ecological aspects, but there are differences in terms of economy and operation. Based on the analysis results of this study, to hold a domestic garden expo in the future, it is necessary to properly utilize "environmental" and "ecological" garden aspects that have potential values according to the region's characteristics to develop sustainable, eco-friendly tourism resources. In addition, values will be more apparent when cultural and artistic programs are planned to establish a differentiated identity in the host area and are appropriately used as a marketing means for a local fair. A well-planned local festival through communication with local residents can affect the image of the region and lead to the revitalization of the local communities by securing urban competitiveness along with the establishment of urban brands, so it can be said that local residents' participation and national or local organizations' cooperation is essential.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Radiation Oncology Journal
• /
• v.12 no.2
• /
• pp.225-232
• /
• 1994

Purpose : This study was to obtain the basic dosimetric data using the 10 MV X-ray for the total body irradiation. Materials and Methods : A linear accelerator photon beam is planned to be used as a radiation source for total body irradiation (TBI) in Chonnam University Hospital. The planned distance from the target to the midplane of a patient is 360cm and the maximum geometric field size is 144cm x 144cm. Polystyrene phantom sized $30{\times}30{\times}30.2cm^3$ and consisted of several sheets with various thickness, and a parallel plate ionization chamber were used to measure surface dose and percent depth dose (PDD) at 345cm SSD, and dose profiles. To evaluate whether a beam modifier is necessary for TBI, dosimetry in build up region was made first with no modifier and next with an 1cm thick acryl plate 20cm far from the polystyrene phantom surface. For a fixed sourec-chamber distance, output factors were measured for various depth. Results : As any beam modifier was not on the way of radiation of 10MV X-ray, the $d_{max}$ and surface dose was 1.8cm and $61\%$, respectively, for 345cm SSD. When an 1cm thick acryl plate was put 20cm far from polystyrene phantom for the SSD, the $d_{max}$ and surface dose were 0.8cm and $94\%$, respectively. With acryl as a beam spoiler, the PDD at 10cm depth was $78.4\%$ and exit dose was a little higher than expected dose at interface of exit surface. For two-opposing fields for a 30cm phantom thick phantom, the surface dose and maximum dose relative to mid-depth dose in our experiments were $102.5\%$ and $106.3\%$, respectively. The off-axis distance of that point of $95\%$ of beam axis dose were 70cm on principal axis and 80cm on diagonal axis. Conclusion: 1. To increase surface dose for TBI by 10MV X-ray at 360cm SAD, 1cm thick acrylic spoiler was sufficient when distance from phantom surface to spoiler was 20cm. 2. At 345cm SSD, 10MV X-ray beam of full field produced a satisfiable dose uniformity for TBI within $7\%$ in the phantom of 30cm thickness by two-opposing irradiation technique. 3. The uniform dose distribution region was 67cm on principal axis of the beam and 80cm on diagonal axis from beam axis. 4. The output factors at mid-point of various thickness revealed linear relation with depth, and it could be applicable to practical TBI.

• The Journal of Korean Society for Radiation Therapy
• /
• v.23 no.2
• /
• pp.83-90
• /
• 2011

Purpose: This study statistically analyzed the difference of the stability of maintaining a respiratory period shown according to position and use of a device to search the tendency and usefulness of a device. Materials and Methods: The study obtained respiratory signals which maintained a respiratory period for 20 minutes each supine and prone position for 11 subjects. The study obtained respiratory signals in a state of using a belly board for 7 patients in a bad condition of a respiratory period in a prone position to analyze a change in respiration and the stability before and after the use of a device. Results: The supine part showed 54.5%, better than the prone part of 36.4% in a case that the stability for maintaining a respiratory period was in a good condition as a fixed respiratory period was well maintained according to the position. 6 patients (85%) showed a maintenance pattern of a respiratory period significantly different before the use and 4 patients showed a significantly good change in the stability for maintaining a respiratory period as a result that belly boards were used for 7 patients that the maintenance of a respiratory period was not in a good condition on a prone position. Conclusion: It seemed that this study could contribute to the maintenance of respiratory period and of respiratory stability as the optimal position for maintenance of respiration and the use of a device such as a belly board were decided through statistic analysis of respiratory signals and its application even if patient position and use of device were decided by the beam arrangement a treatment part of a patient, location of a target, and an expected plan.

• The Journal of Korean Society for Radiation Therapy
• /
• v.19 no.1
• /
• pp.27-33
• /
• 2007

Purpose: We have performed SRS (stereotactic radiosurgery) for avm (arterry vein malformation) and brain cancer. In order to verify dose and localization of SRS, dose distributions from TPS ($X-Knife^{(R)}$ 3.0, Radionics, USA) and GafChromic $EBT^{(R)}$ film in a head phantom were compared. Materials and Methods: In this study, head and neck region of conventional humanoid phantom was modified by substituting one of 2.5 cm slap with five 0.5 cm acrylic plates to stack the GafChromic $EBT^{(R)}$ film slice by slice with 5 mm intervals. Four films and five acrylic plates were cut along the contour of head phantom in axial plane. The head phantom was fixed with SRS head ring and adapted SRS localizer as same as real SRS procedure. CT images of the head phantom were acquired in 5 mm slice intervals as film interval. Five arc 6 MV photon beams using the SRS cone with 2 cm diameter were delivered 300 cGy to the target in the phantom. Ten small pieces of the film were exposed to 0, 50, 100, 200, 300, 400, 500, 600, 700, 800, 900 cGy, respectively to calibrate the GafChromic $EBT^{(R)}$ film. The films in the phantom were digitized after 24 hours and its linearity was calibrated. The pixel values of the film were converted to the dose and compared with the dose distribution from the TPS calculation. Results: Calibration curve for the GafChromic $EBT^{(R)}$ film was linear up to 900 cGy. The R2 value was better than 0.992. Discrepancy between calculated from $X-Knife^{(R)}$ 3.0 and measured dose distributions with the film was less than 5% through all slices. Conclusion: It was possible to evaluate every slice of humanoid phantom by stacking the GafChromic EBT film which is suitable for 2 dimensional dosimetry, It was found that film dosimetry using the GafChromic $EBT^{(R)}$ film is feasible for routine dosimetric QA of stereotactic radiosurgery.

• Progress in Medical Physics
• /
• v.25 no.1
• /
• pp.15-22
• /
• 2014

The IMRT is proper implement to get high dose deliver to tumor as its shape and selective approach in radiation therapy. Since the IMRT is performed as modulated the radiation fluence by the MLC created the open shapes and its irradiation time, the dose of segment of radiation field effects on the cumulated portal dose. The accurate output factor of small and step shape of segment is important to improve the determination of deliver tumor dose as it is directly proportional to dose. This experiment performed with the 6 MV photon beam of Clinac Ex(Varian) from $3{\times}3cm^2$ to $0.5{\times}0.5cm^2$ small field size for collimator jaw in MLC free and/or for MLC open field in fixed collimator jaw $10{\times}10cm^2$ using the CC01 ion chamber, SFD diode, diamond detector and X-Omat film dosimetry. As results of normalized to the reference field of $10{\times}10cm^2$ of MLC, the output factor of $3{\times}3cm^2$ showed $0.899{\pm}0.0106$, $0.855{\pm}0.0106$ for $2{\times}2cm^2$, $0.764{\pm}0.0082$ for $1{\times}1cm^2$ and $0.602{\pm}0.0399$ for $0.5{\times}0.5cm^2$. The output factor of MLC open field has shown a maximum 3.8% higher than that of the collimator jaw open field.

• The Korean Journal of Pesticide Science
• /
• v.11 no.3
• /
• pp.201-209
• /
• 2007

Characteristics of the 5 biopesticides that included Bacillus thuringiensis and on the domestic markets were investigated. These products were contained different strains of B. thuringiensis, for examples; product A and E was B. thuringiensis subsp aizawai; product B was B. thuringiensis; product C was B. thuringiensis Berline var. kurstaki; product D was B. thuringiensis var. kurstaki. Number of active spores were counted because they could influence the bio-activity against target pests. Only product C are contained the fixed quantity as its label, however, product D and E were a tenth part, and product A and B were a hundredth part of their descriptions. The pHs of product A and B were measured 3.67 and 3.73, and C, D and E were 5, respectively. Typical bypyramidal crystals produced from B. thuringiensis was found in only product C under a phase contrast microscope. For the uniform formulation of products that conformed whether B. thuringiensis were equally spreaded on the crops, B. thuringiensis in the C, D and E were equally grown on the nutrient agar medium As a results, product A were more different from product C than any other products. When product A and C were bioassayed against different larval stages of diamondback moth, their mortalities with spraying application were showed 100% after 48 hours.

• Korean Journal of Soil Science and Fertilizer
• /
• v.41 no.2
• /
• pp.126-136
• /
• 2008

Tobacco plants grown in pots by sand culture for 70 days after transplanting were used to evaluate the sensing distance and measurement efficiency of ground-based remote sensors. The leaf distribution of tobacco plant and sensing distance from the sensors to the target leaves were controlled by two removal methods of leaves, top-down and bottom-up removal. In the case of top-down removal, the canopy reflectance was measured by the sensor located at a fixed position having an optimum distance from the detector to the uppermost leaf of tobacco every time that the higher leaves were one at a time. The measurement of bottom-up removal, a the other hand, was conducted in the same manner as that of the top-down removal except that the lower leaves were removed one by one. Canopy reflectance measurements were made with hand held spectral sensors including the active sensors such as $GreenSeeker^{TM}$ red and green, $Crop\;Circle\;ACS-210^{TM}$ red and amber, the passive sensors of $Crop\:Circle^{TM}$, and spectroradiometer $SD2000^{TM}$. The reflectance indices by all sensors were generally affected by the upper canopy condition rather than lower canopy condition of tobacco regardless of sensor type, passive or active. The reflectance measurement by $GreenSeeker^{TM}$ was affected sensitively at measurement distance longer than 120 cm, the upper limit of effective sensing distance, beyond which measurement errors are appreciable. In case of the passive sensors that has no upper limit of effective distance and $Crop\;Circle^{TM}(ACS210)$ that has the upper limit of effective sensing distance specified with 213 cm, longer than that of estimated distance, the measurement efficiency affected by the sensing distance showed no difference. This result suggests that it is necessary to use the sensor specified optimum distance. The result revealed that active sensors are more superior than their passive counterparts in establishing between the relative ratio of reflectance index and the dry weight of tobacco treated by top-down removal, and in the evaluation of biomass. $The\;Crop\;Circle\;ACS-210^{TM}$ red was proved to have the highest efficiency of measurement, followed by $Crop\;Circle^{TM}(ACS210)$ amber and $GreenSeeker^{TM}$ red, $Crop\;Circle^{TM}$ passive, $GreenSeeker^{TM}$ green, and spectroradiometer, in descending order.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.50 no.2
• /
• pp.41-63
• /
• 2022

The purpose of this study is to secure objective and precise data through ecosystem monitoring, to reveal ecological characteristics through comparison and analysis with past survey data, and to accumulate basic data for diagnosing the current situation and predicting changes in the ecosystem. The target site is the 'Quercus mongolica forest on the Buksa-myeon of Namsan', which was designated as an Ecological Landscape Conservation Area (ELCA) of Seoul in July 2006. The research contents are analysis of soil environment change (1986~2016), change of actual vegetation (1978~2016), and change of plant community structure (1994~2016). A total of 8 fixed surveys (400~1,200m²) were established in 1994 and 2000. Analysis items are importance value, species and population, and Shannon's species diversity. The soil environment of Namsan is acidic (pH 4.40 in 2016), which is expected to have a negative impact on tree growth and vegetation structure due to its low capacity for exchangeable cations. Quercus mongolica forest in Namsan is mainly distributed on the northern slopes. The actual vegetation area changed from 49.4% in 1978 → 80.7% in 1986 → 82.4% in 2000 → 88.3% in 2005 → 88.3% in 2009 → 70.3% in 2016. In 2016, the forest decreased by 18% compared to 2009. While there was increased growth of Quercus mongolica in the tree layer from 2009 to 2016, the overall decline in vegetation area was due to logging and fumigation management following the spread of oak wilt in 2012. As for the changes in the plant community structure, Quercus mongolica of the tree layer was damaged by oak wilt, and the potential vegetation that can form the next generation was ambiguous. In the subtree layer, the force of urbanization tree species such as Styrax japonicus, Sorbus alnifolia, and Acer palmatum. was maintained or increased. In the shrub layer, the number of trees and species increased significantly due to the open tree crown, and accordingly, the species diversity of Shannon for woody plants also increased. In Quercus mongolica forest of Namsan, various ecological changes are occurring due to the effects of urban environments such as air pollution and acid rain, the limitation of Quercus mongolica pure forest due to oak wilt, and the introduction of exotic species, thus, it is necessary to establish a management plan through continuous monitoring.

• Journal of Distribution Research
• /
• v.17 no.1
• /
• pp.1-36
• /
• 2012

Service failure is one of the major reasons for customer defection. As the business environment gets tougher and more competitive, a single service failure might bring about fatal consequences to a service provider or a firm. Sometimes a failure won't end up with an unsatisfied customer's simple complaining but with a wide-spread animosity against the service provider or the firm, leading to a threat to the firm's survival itself in the society. Therefore, we are in need of comprehensive understandings of complainants' attitudes and behaviors toward service failures and firm's recovery efforts. Even though a failure itself couldn't be fixed completely, marketers should repair the mind and heart of unsatisfied customers, which can be regarded as an successful recovery strategy in the end. As the outcome of recovery efforts exerted by service providers or firms, recovery of the relationship between customer and service provider need to put on the top in the recovery goal list. With these motivations, the study investigates how service failure and recovery makes the changes in dynamics of fundamental elements of customer-firm relationship, such as customer affection, customer trust and loyalty intention by comparing two time points, before the service failure and after the recovery, focusing on the effects of recovery satisfaction and the failure severity. We adopted La & Choi (2012)'s framework for development of the research model that was based on the previous research stream like Yim et al. (2008) and Thomson et al. (2005). The pivotal background theories of the model are mainly from relationship marketing and social relationships of social psychology. For example, Love, Emotional attachment, Intimacy, and Equity theories regarding human relationships were reviewed. As the results, when recovery satisfaction is high, customer affection and customer trust that were established before the service failure are carried over to the future after the recovery. However, when recovery satisfaction is low, customer-firm relationship that had already established in the past are not carried over but broken up. Regardless of the degree of recovery satisfaction, once a failure occurs loyalty intention is not carried over to the future and the impact of customer trust on loyalty intention becomes stronger. Such changes imply that customers become more prudent and more risk-aversive than the time prior to service failure. The impact of severity of failure on customer affection and customer trust matters only when recovery satisfaction is low. When recovery satisfaction is high, customer affection and customer trust become severity-proof. Interestingly, regardless of the degree of recovery satisfaction, failure severity has a significant negative influence on loyalty intention. Loyalty intention is the most fragile target when a service failure occurs no matter how severe the failure criticality is. Consequently, the ultimate goal of service recovery should be the restoration of customer-firm relationship and recovery of customer trust should be the primary objective to accomplish for a successful recovery performance. Especially when failure severity is high, service recovery should be perceived highly satisfied by the complainants because failure severity matters more when recovery satisfaction is low. Marketers can implement recovery strategies to enhance emotional appeals as well as fair treatments since the both impacts of affection and trust on loyalty intention are significant. In the case of high severity of failure, recovery efforts should be exerted to overreach customer expectation, designed to directly repair customer trust and elaborately designed in the focus of customer-firm communications during the interactional recovery process to affect customer trust rebuilding indirectly. Because it is a longer and harder way to rebuild customer-firm relationship for high severity cases, low recovery satisfaction cannot guarantee customer retention. To prevent customer defection due to service failure of high severity, unexpected rewards as a recovery will be likely to be useful since those will lead to customer delight or customer gratitude toward the service firm. Based on the results of analyses, theoretical and managerial implications are presented. Limitations and future research ideas are also discussed.