• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.35S no.5
• /
• pp.83-93
• /
• 1998

Three dimensional data acquisition system based on the structured light is developed in this work. The system is composed of a CCD camera, slide projector, and various image processing programs. Calibration procedures and several image processing steps which are necessary to get the rnage data are described. A new grid labeling technique and a grid pattern are devised to improve the accuracy of th eobtained data. Preliminary experimental result shows that the developed system may be used as a simple and cheap 3D data acquisition system. Severla suggestions are included for further research.

• Journal of Institute of Convergence Technology
• /
• v.11 no.1
• /
• pp.1-5
• /
• 2021

Deep learning based methods achieve state-of-the-art accuracy, however, they typically rely on supervised training with large labeled datasets. It is known in many medical applications that labeling medical images requires significant expertise and much time, and typical hand-tuned approaches for data augmentation fail to capture the complex variations in such images. This paper proposes a 3D image augmentation method to overcome these difficulties. It allows us to enrich diversity of training data samples that is essential in medical image segmentation tasks, thus reducing the data overfitting problem caused by the fact the scale of medical image dataset is typically smaller. Our numerical experiments demonstrate that the proposed approach provides significant improvements over state-of-the-art methods for 3D medical image segmentation.

• The Korean Journal of Community Living Science
• /
• v.22 no.3
• /
• pp.379-393
• /
• 2011

This study compared the college students' dietary behavior for processed foods, who will be the main consumers in the future and looked for the measures to understand and establish the right food labeling system by surveying the level of understanding and utilization of food labeling. The data was analysed by SPSS win 17.0 program, and the results are as follows. For the standard of selecting processed foods, the group with high nutrition knowledge considered the reliability of foods as important and the group with low nutrition knowledge considered the products introduced in TV commercial as important. When purchasing processed foods, the group with high nutrition knowledge considered nutrition, taste, price, appearance(shape), and the consumable period more than the group with low nutrition knowledge. For trans fat, the group with high nutrition knowledge learned more about it than the group with low nutrition knowledge. The ratio of confirming food nutrition label was higher in the group with high nutrition knowledge. Regarding the level of confirming individual food labels, the highest level was for milk and dairy products. And there was significant difference for the processed products of meat, cookies, bread and noodles. It was found that the level of confirmation was higher in the group with high nutrition knowledge. And the most important indication for individual food product was the consumable period. To preserve the purchased foods, the group with high nutrition knowledge preserve the foods in line with the description written on the food cover sheet, and this group used to return or exchange the products when they found them spoiled or purchased by mistake. The group with high nutrition knowledge knew more about the nutrition indication than the group with low nutrition knowledge. The necessity of nutrition indication for processed foods and the need of education and PR(Public Relation) were acknowledged higher in the group with high nutrition knowledge. For the effect of nutrition indication, it showed that the group with high nutrition knowledge thought it would improve the quality and the group with low nutrition knowledge thought it would be helpful when comparing the product with others. The group with high nutrition knowledge showed higher understanding level about nutrition indication than the group with low nutrition knowledge.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.366-369
• /
• 2021

본 논문에서는 동영상에서 일련의 과정을 거쳐 얻었던 학습데이터를 보다 간편하고 빠른 속도로 획득하는 방법을 제안한다. 음성과 영상 스트림을 처리하는 ffmpeg을 이용해 영상을 프레임화하고, 딥 러닝 기반의 YOLO 알고리즘을 사용하여 객체를 검출한다.

• Journal of Platform Technology
• /
• v.6 no.1
• /
• pp.3-8
• /
• 2018

This paper proposes a technique for recognizing online handwritten cursive data obtained by tracing a motion trajectory while a user is in the 3D space based on a convolution neural network (CNN) algorithm. There is a difficulty in recognizing the virtual character input by the user in the 3D space because it includes both the character stroke and the movement stroke. In this paper, we divide syllable into consonant and vowel units by using labeling technique in addition to the result of localizing letter stroke and movement stroke in the previous study. The coordinate information of the separated consonants and vowels are converted into image data, and Korean handwriting recognition was performed using a convolutional neural network. After learning the neural network using 1,680 syllables written by five hand writers, the accuracy is calculated by using the new hand writers who did not participate in the writing of training data. The accuracy of phoneme-based recognition is 98.9% based on convolutional neural network. The proposed method has the advantage of drastically reducing learning data compared to syllable-based learning.

• Journal of the Korea Society of Computer and Information
• /
• v.23 no.9
• /
• pp.81-86
• /
• 2018

Software defect severity is very important in projects with limited historical data or new projects. But general software defect prediction is very difficult to collect the label information of the training set and cross-project defect prediction must have a lot of data. In this paper, an unclassified data set with defect severity is clustered according to the distribution ratio. And defect severity-based prediction model is proposed by way of labeling. Proposed model is applied CLAMI in JM1, PC4 with the least ambiguity of defect severity-based NASA dataset. And it is evaluated the value of ACC compared to original data. In this study experiment result, proposed model is improved JM1 0.15 (15%), PC4 0.12(12%) than existing defect severity-based prediction models.

• Journal of the Korea Society of Computer and Information
• /
• v.24 no.9
• /
• pp.21-27
• /
• 2019

In this paper, we propose a deep auto-encoder-based pipe leak detection (PLD) technique from time-series acoustic data collected by microphone sensor nodes. The key idea of the proposed technique is to learn representative features of the leak-free state using leak-free time-series acoustic data and the deep auto-encoder. The proposed technique can be used to create a PLD model that detects leaks in the pipeline in an unsupervised learning manner. This means that we only use leak-free data without labeling while training the deep auto-encoder. In addition, when compared to the previous supervised learning-based PLD method that uses image features, this technique does not require complex preprocessing of time-series acoustic data owing to the unsupervised feature extraction scheme. The experimental results show that the proposed PLD method using the deep auto-encoder can provide reliable PLD accuracy even considering unsupervised learning-based feature extraction.

• Journal of Internet Computing and Services
• /
• v.24 no.6
• /
• pp.91-98
• /
• 2023

Recently, according to development of artificial intelligence, a wide range of industry being automatic and optimized. Also we can find out some research of using supervised learning for deteceting defect of railway in domestic rail industry. However, there are structures other than rails on the track, and the fastener is a device that binds the rail to other structures, and periodic inspections are required to prevent safety accidents. In this paper, we present a method of reducing cost for labeling using semi-supervised and transfer model trained on rail fastener data. We use Resnet50 as the backbone network pretrained on ImageNet. At first we randomly take training data from unlabeled data and then labeled that data to train model. After predict unlabeled data by trained model, we adopted a method of adding the data with the highest probability for each class to the training data by a predetermined size. Futhermore, we also conducted some experiments to investigate the influence of the number of initially labeled data. As a result of the experiment, model reaches 92% accuracy which has a performance difference of around 5% compared to supervised learning. This is expected to improve the performance of the classifier by using relatively few labels without additional labeling processes through the proposed method.

• Journal of the Korean Society of Food Culture
• /
• v.12 no.5
• /
• pp.573-582
• /
• 1997

The purpose of this study is to establish the one serving size of Korean Processed Food. Defining the one serving size is very important for nutrition labeling and foodservice operation, because the one serving size is used to set up a proper portion by each foodservice operation. The basic data of 200 items were collected through three methods. Searching many cookbooks, exploring the commercial and noncommercial foodservices -6 industrial foodservices, 100 nationwide elementary school foodservice recipes analysis, and 3 hospital foodservice systems as the samples - moreover, experimental cooking and sensory evaluation by trained panels were conducted to assess quantity preference of selected food items. All data were rearranged through food type, that is, main dish, side dish, dessert and health food. One serving sizes of processed foods showed wide variety according to the different menus that include selected food items. Therefore, means and ranges of serving size by three research methods were presented item by item. The results obtained were: 1. The Korean Processed Foods were dried and sugar adding and soused foods, and many of them used the natual processing methods. 2. There were wide varieties in the classification of main dishes, but many of them were cereals, noodles, and sugar products. One serving size of noodles were around $50{\sim}100\;g$, cereals were $20{\sim}40\;g$, which means the one serving size can be differenciated by the food usage. 3. According to the Food classification of side dishes, many of them were as following; natural dried foods, processed fish products, salted or sugar added foods, seasoned foods and sugar products. Moreover the Types of cooking in side dishes were almost culinary vegetables, teas, health foods and condiments, and soused fish products. 4. About desserts, they were almost teas and sugars, and the Types of cooking were teas, health foods and seasonings. 5. We can conclude that almost Korean Processed foods used the drying and soused processing methods for long-time preservation, but it can make the higher content of any special elements, such as sodium or carbohydrates.

• KIPS Transactions on Software and Data Engineering
• /
• v.6 no.9
• /
• pp.445-456
• /
• 2017

In this paper, we propose a method to detect spam tweets containing unhealthy information by using an n-gram dictionary under limited labeling. Spam tweets that contain unhealthy information have a tendency to use similar words and sentences. Based on this characteristic, we show that spam tweets can be effectively detected by applying a Naive Bayesian classifier using n-gram dictionaries which are constructed from spam tweets and normal tweets. On the other hand, constructing an initial training set requires very high cost because a large amount of data flows in real time in a twitter. Therefore, there is a need for a spam detection method that can be applied in an environment where the initial training set is very small or non exist. To solve the problem, we propose a method to generate pseudo-labels by utilizing twitter's retweet function and use them for the configuration of the initial training set and the n-gram dictionary update. The results from various experiments using 1.3 million korean tweets collected from December 1, 2016 to December 7, 2016 prove that the proposed method has superior performance than the compared spam detection methods.