• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.7
• /
• pp.2390-2406
• /
• 2022

Scene text recognition has important application value and attracted the interest of plenty of researchers. At present, many methods have achieved good results, but most of the existing approaches attempt to improve the performance of scene text recognition from the image level. They have a good effect on reading regular scene texts. However, there are still many obstacles to recognizing text on low-quality images such as curved, occlusion, and blur. This exacerbates the difficulty of feature extraction because the image quality is uneven. In addition, the results of model testing are highly dependent on training data, so there is still room for improvement in scene text recognition methods. In this work, we present a natural scene text recognizer to improve the recognition performance from the feature level, which contains feature representation and feature enhancement. In terms of feature representation, we propose an efficient feature extractor combined with Representative Batch Normalization and ResNet. It reduces the dependence of the model on training data and improves the feature representation ability of different instances. In terms of feature enhancement, we use a feature enhancement network to expand the receptive field of feature maps, so that feature maps contain rich feature information. Enhanced feature representation capability helps to improve the recognition performance of the model. We conducted experiments on 7 benchmarks, which shows that this method is highly competitive in recognizing both regular and irregular texts. The method achieved top1 recognition accuracy on four benchmarks of IC03, IC13, IC15, and SVTP.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.35 no.5
• /
• pp.259-266
• /
• 2022

In this study, we developed a deep learning and vision sensor-based assembly performance evaluation method isfor prefabricated steel structures. The assembly parts were segmented using a modified version of the receptive field block convolution module inspired by the eccentric function of the human visual system. The quality of the assembly was evaluated by detecting the bolt holes in the segmented assembly part and calculating the bolt hole positions. To validate the performance of the evaluation, models of standard and defective assembly parts were produced using a 3D printer. The assembly part segmentation network was trained based on the 3D model images captured from a vision sensor. The sbolt hole positions in the segmented assembly image were calculated using image processing techniques, and the assembly performance evaluation using the k-nearest neighbor algorithm was verified. The experimental results show that the assembly parts were segmented with high precision, and the assembly performance based on the positions of the bolt holes in the detected assembly part was evaluated with a classification error of less than 5%.

• The Journal of Korea Robotics Society
• /
• v.17 no.3
• /
• pp.365-372
• /
• 2022

The accurate object recognition is important for the precise and accurate manipulation. To enhance the recognition performance, we can use various types of sensors. In general, acquired data from sensors have a high sampling rate. So, in the past, the RNN-based model is commonly used to handle and analyze the time-series sensor data. However, the RNN-based model has limitations of excessive parameters. CNN-based model also can be used to analyze time-series input data. However, CNN-based model also has limitations of the small receptive field in early layers. For this reason, when we use a CNN-based model, model architecture should be deeper and heavier to extract useful global features. Thus, traditional methods like RN N -based and CN N -based model needs huge amount of learning parameters. Recently studied result shows that Fast Fourier Convolution (FFC) can overcome the limitations of traditional methods. This operator can extract global features from the first hidden layer, so it can be effectively used for feature extracting of sensor data that have a high sampling rate. In this paper, we propose the algorithm to recognize objects using tactile sensor data and the FFC model. The data was acquired from 11 types of objects to verify our posed model. We collected pressure, current, position data when the gripper grasps the objects by random force. As a result, the accuracy is enhanced from 84.66% to 91.43% when we use the proposed FFC-based model instead of the traditional model.

• Journal of Korean Academy of Nursing
• /
• v.54 no.3
• /
• pp.459-474
• /
• 2024

Purpose: The purpose of this study was to conduct a concept analysis of social intelligence in nurses so that applying social intelligence to the nursing field. Methods: In this study, we followed the hybrid model procedure, involving the following steps: First, in the theoretical stage, the attributes and definitions of the concept of social intelligence were determined through literature review. Second, the concepts' reality was confirmed during fieldwork. In the final analysis stage, the results confirmed in the theoretical and fieldwork stages were compared and analyzed to confirm the properties and definition of the concept. Results: Nurses' social intelligence consists of three dimensions: social cognitive nursing competency, human-centered social evolution, and skills for solving complex nursing situations. Nurses' social intelligence is a professional nursing competency that flexibly coordinates complex nursing situations, developed through accumulating experiences of continuous reflection and relationship expansion based on receptive listening and social sensitivity in clinical interpersonal relationships. Conclusion: Nurses' social intelligence is widely used in clinical practice and is shown to have a significant direct and indirect impact on clinical nursing. To effectively apply social intelligence in the clinical context, individual and organizational efforts are required to share and transfer knowledge and capacity-building methods through collective intelligence and education.

• Korean Journal of Veterinary Research
• /
• v.37 no.1
• /
• pp.119-128
• /
• 1997

The relaxation of gastric fundus smooth muscles is the primary physiological event which induces the receptive relaxation of monogastric animals. L-arginine/Nitric oxide(L-arg/NO) system is known to mediate the inhibitory non-adrenergic non-cholinergic(NANC) neurotransmission in various tissues including gastrointestinal smooth muscles. The longitudinal smooth muscles of porcine gastric fundus showed fast relaxation during electrical field stimulation(EFS) and rebound contraction after EFS in NANC condition. So, the purpose of present study was elucidation of the neurotrasmitters related to the NANC relaxation and explanation of the relation between NANC relaxation and L-arg/NO system. The longitdinal smooth muscles of porcine gastric fundus were hung in the organ bath and under the presence of guanethidine($5{\times}10^{-5}M$), precontraction was induced by carbachol($1{\times}10^{-6}M$). The muscle responses to EFS and drugs were isomerically recorded. The rusults were summarized as follows. 1. The longtudinal muscles of porcine gastric fundus showed frequency-dependent relaxation and rebound contraction to electrical field stimulaton(1ms, 8V, 1~16Hz, 20sec, EFS). These responses were blocked by tetrodotoxin($1{\times}10^{-6}M$). 2. The relaxation and rebound contraction of the longitudinal muscles of porcine gastric fundus to EFS were inhibited by L-NAME($2{\times}10^{-5}M$). The inhibitory effect of L-NAME was antagonized by L-arginine($1{\times}10^{-3}M$), but not by D-arginine($1{\times}10^{-3}M$). 3. Exogenous NO($NaNO_2$, $1{\times}10^{-5}{\sim}1{\times}10^{-4}M$, pH=2.0) caused concentration-dependent relaxation as EFS did. 4. Methylene Blue($2{\times}10^{-5}M$), a soluble guanylate cyclase inhibitor, inhibited the relaxation and rebound contraction of the longitudinal muscles of porcine gastric fundus induced by EFS, but N-ethlmaleimide, a adenylate cyclase inhibitor, did not. 5. 8-Br-cGMP($1{\times}10^{-6}{\sim}3{\times}10^{-6}M$), permeable cGMP analogue, induced dose-dependent relaxation. but 8-Br-cAMP($1{\times}10^{-6}{\sim}3{\times}10^{-6}M$), permeable cAMP analogue, did not. Both did not evoked rebound contraction. 6. ${\alpha}$-chymotrypsin did not affect the relaxation of the longitudinal muscles of porcine gastric fundus. 7. Reactive blue 2($1{\times}10^{-4}M$, 40min) siginificantly inhibited the rebound contraction induced by EFS and inhibited contraction caused by exogenous ATP($1{\times}10^{-4}{\sim}1{\times}10^{-3}M$). These results suggests that NANC relaxation of the longitudinal muscles of porcine gastric fundus mainly mediated by NO and the rebound contraction is related to NO and other neurotransmitters.

• Journal of the Korean Institute of Traditional Landscape Architecture
• /
• v.31 no.3
• /
• pp.75-82
• /
• 2013

This study is a basic study to develop standards and foundations for the establishment of traditional Korean gardens and aims to understand the current status of their components and expression methods and identify problems by investigating Korean gardens developed overseas. Nine sites were selected for field surveys and monitoring assessments. The results suggest: Overall, there is a lack of popular generality and temporal characteristics among these gardens, as they are mere reproductions of historical elements. There have also been errors of traditional and experimental interpretations. In terms of design aspects, traditional gardens are primarily compilations of landscape elements and certain ornamental features. In terms of landscape, they tend to be insufficient in parlaying appropriate spatial scales and experiential hierarchies; they also lack considerations of the context of neighbouring landscapes. In terms of guidance and information delivery, there is a worldwide lack, in general, of recognition of Korean gardens, given the broad variety of names attached to them; therefore, name standardization is recommended. In terms of development, management, and use, it is essential that designers suggest plant types, as well as alternatives, that match the characteristics of a given space; a receptive attitude vis-$\grave{a}$-vis the characteristics of their use is required.

• Journal of Sensor Science and Technology
• /
• v.11 no.2
• /
• pp.84-92
• /
• 2002

A $32{\times}32$ pixels foveated (linear-polar) structure retina chip with the function of local-light adaptation for edge detection has been designed and fabricated using CMOS technology. Human retina can detect a wide range of light intensity. In this study, we use the biologically-inspired visual signal processing mechanism that consists of photoreceptors, horizontal cells, and bipolar cells in order to implement the function of edge detection in the retina chip. For a local-light adaptive function, the size of receptive field is changed locally according to the input light intensity. The spatial distribution of sensing pixels in the foveated retina chip has the advantages of selective reduction of image data and good resolution in central part to carry out the elaborate image processing with still enough resolution in the outer parts. The designed chip has been fabricated using standard $0.6\;{\mu}m$ double-poly triple-metal CMOS technology and optimized using HSPICE simulator.

• Journal of Music and Human Behavior
• /
• v.10 no.1
• /
• pp.1-23
• /
• 2013

Upper extremity dysfunction is a common consequence following stroke. Spontaneous recovery during the first six months post-stroke is rigorous and considered as a significant indicator of potential long-term progress. Various approaches have been utilized to regain functional upper limb movement necessary for independent living; however, conventional therapy approaches have failed to prove consistency, especially for subacute stroke patients. There is, thus, a need for innovative therapeutic strategies that motivate stroke survivors to facilitate neural and functional recovery during the critical window immediately following stroke. The effect of music on physical enhancement has been frequently reported in the field of medicine as well as neurorehabilitation. The efficacy of rhythm on lower extremity deficits has been well established. Yet, the rationale for using instrumental music making enhancing subacute upper extremities rehabilitation is not clearly described to date. Based on the key mechanism of music as sensori-motor movement facilitator, this paper reviews previous empirical research that utilized music-based interventions for upper extremity rehabilitation for stroke patients, either in the form of receptive or expressive activity. This paper, further, focuses on the current research trends in subacute stroke upper limb rehabilitation and provides applicable rationale of using instrumental music playing.

• Journal of Broadcast Engineering
• /
• v.26 no.4
• /
• pp.429-440
• /
• 2021

With the recent development of deep convolutional neural network learning, deep learning techniques applied to single image super-resolution are showing good results. One of the existing deep learning-based super-resolution techniques is RDN(Residual Dense Network), in which the initial feature information is transmitted to the last layer using residual dense blocks, and subsequent layers are restored using input information of previous layers. However, if all hierarchical features are connected and learned and a large number of residual dense blocks are stacked, despite good performance, a large number of parameters and huge computational load are needed, so it takes a lot of time to learn a network and a slow processing speed, and it is not applicable to a mobile system. In this paper, we use the residual dense structure, which is a continuous memory structure that reuses previous information, and the residual dense channel attention block using the channel attention method that determines the importance according to the feature map of the image. We propose a method that can increase the depth to obtain a large receptive field and maintain a concise model at the same time. As a result of the experiment, the proposed network obtained PSNR as low as 0.205dB on average at 4× magnification compared to RDN, but about 1.8 times faster processing speed, about 10 times less number of parameters and about 1.74 times less computation.

• The Korean Journal of Physiology
• /
• v.24 no.1
• /
• pp.181-194
• /
• 1990

Antidromically activated spinoreticular tract (SRT) cell units in the lumbosacral enlargement of ${\alpha}-chloralose$ anesthetized cats were classified as medial and lateral SRT units according to the location of their axonal termination. Identified SRT units were tested fer antidromic conduction velocity, laterality of their axonal projection, the location in spinal gray, peripheral receptive field, the response pattern to graded mechanichal stimulation and the responsiveness to $A{\delta}$ and C volley of the peripheral nerve. 1) The 59% of 34 medial SRT units were recorded in ipsilateral side to the antidromic stimulation site, but 60% of the 47 lateral SRT units projected to contralateral side. 2) Most of the medial SRT cells and rostral ventrolateral medulla (RVLM)-projecting lateral SRT cells were recorded in lamina VII & VIII. The LRN (lateral reticular nucleus)-projecting SRT cells, however, distributed through all the laminae except superficial ones (I & II). 3) The identified SRT units were classified as low theshold (LT), deep, high threshold (HT), wide dynamic range (WDR) cells, based on the response patterns to graded mechanical stimuli. The proportion of SRT units which receive noxious input was 37.5%, 25% and 75% in the medial, LRN-projecting and RVLM SRT group, respectively. 4) There was no significant difference in the mean conduction velocities between the 3 groups. But the deep cells had significantly higher velocity than that of the HT cells. The above results show that the peripheral inputs to the SRT units are different in the 3 groups: medial, LRN & RVLM SRT group. Especially in case of the SRT cells projecting to RVLM which is a probable candidate fur the integration center of various pressor reflexes such as somatosympathetic reflex, the noxious informations occupy higher proportion of input to them than in other groups. Therefore the noxious information transmitted through the lateral SRT destined for RVLM is expected to play a role in somatosymapthetic reflex.