• 말소리와 음성과학
• /
• 제15권2호
• /
• pp.13-20
• /
• 2023

This study explores the potential of automated speech recognition (ASR) in assessing English learners' pronunciation. We employed ASR technology, acknowledged for its impartiality and consistent results, to analyze speech audio files, including synthesized speech, both native-like English and Korean-accented English, and speech recordings from a native English speaker. Through this analysis, we establish baseline values for the word error rate (WER). These were then compared with those obtained for human raters in perception experiments that assessed the speech productions of 30 first-year college students before and after taking a pronunciation course. Our sub-group analyses revealed positive training effects for Whisper, an ASR tool, and human raters, and identified distinct human rater strategies in different assessment aspects, such as proficiency, intelligibility, accuracy, and comprehensibility, that were not observed in ASR. Despite such challenges as recognizing accented speech traits, our findings suggest that digital tools such as ASR can streamline the pronunciation assessment process. With ongoing advancements in ASR technology, its potential as not only an assessment aid but also a self-directed learning tool for pronunciation feedback merits further exploration.

• 음성과학
• /
• 제15권3호
• /
• pp.133-144
• /
• 2008

The present study examined multicultural family children's articulation accuracy and phonological process using Assessment of Phonology and Articulation for Children(APAC), and compared them with normally developing children. The subjects of this study were 24 multicultural family children between ages 3 years, 6 months and 3 years, 11 months. The multicultural family children's articulation accuracy was significantly lower than the normally developing children's. In case of the normally developing children, phonological processes the multicultural family children showed were observed at a younger age and did not appear at the age of the children participating in this study. The Japanese multicultural family children and the non Japanese multicultural family children showed different rate of the changes. This result shows that articulation development in the multicultural family children may be different among them according to the classification and that the children's error patterns are related to their mothers' native language. The results of this study are proposed to be applicable to articulation assessment and treatment.

• 환경영향평가
• /
• 제14권3호
• /
• pp.127-134
• /
• 2005

Recently, aquaculture farm sites have been increased with demand of the expensive fish species and sea food like as seaweed, laver and oyster. Therefore coastal water quality have been deteriorated by organic contamination from marine aquaculture farm sites. For protecting of coastal environment, we need to control the location of aquaculture sites. The purpose of this study is to detect the laver aquaculture sites using multispectral remotely sensed data with autodetection algorithm. In order to detect the aquaculture sites, density slice and contour and vegetation index methods were applied with SPOT and IKONOS data of Shinan area. The marine aquaculture farm sites were extracted by density slice and contour methods with one band digital number(DN) carrying 65% accuracy. However, vegetation index algorithm carried out 75% accuracy using near-infra red and red bands. Extraction of the laver aquaculture site using remotely sensed data will provide the efficient digital map for coastal water management strategies and red tide GIS management system.

• 반도체디스플레이기술학회지
• /
• 제19권4호
• /
• pp.126-129
• /
• 2020

We introduce a lifetime assessment technique using deep learning algorithm with complex electrical parameters such as resistivity, permittivity, impedance parameters as integrated indicators for predicting the degradation of the organic molecules. The evaluation system consists of fully automated in-situ measurement system and multiple layer perceptron learning system with five hidden layers and 1011 perceptra in each layer. Prediction accuracies are calculated and compared depending on the physical feature, learning hyperparameters. 62.5% of full time-series data are used for training and its prediction accuracy is estimated as r-square value of 0.99. Remaining 37.5% of the data are used for testing with prediction accuracy of 0.95. With k-fold cross-validation, the stability to the instantaneous changes in the measured data is also improved.

• Smart Structures and Systems
• /
• 제31권1호
• /
• pp.79-88
• /
• 2023

As the number of deteriorated buildings increases, the importance of safety diagnosis, maintenance, and the repair of buildings also increases. Traditionally, building condition assessments are performed by one person or one company and various inspections are needed. This entails a subjective judgment by the inspector, resulting in different assessment results, poor objectivity and a lack of reliability. Therefore, this study proposed a method to bring about accurate grading results of building conditions. The limitations of visual inspection and condition assessment processes previously conducted were identified by reviewing existing studies. Building defect data was collected using the reverse-engineered three-dimensional (3D) model. The accuracy of the results was verified by comparing them with the actual evaluation results. The results show a 50% time-saving to the same area with an accuracy of approximately 90%. Consequently, defect data with high objectivity and reliability were acquired by measuring the length, area, and width. In addition, the proposed method can improve the efficiency of the building diagnosis process.

• 환경영향평가
• /
• 제19권6호
• /
• pp.647-656
• /
• 2010

As the $CO_2$ from buildings take up approximately 25% of the total $CO_2$ emission, the need for regulating and managing this emission is urgently required. Thus this study recognizes $CO_2$ emission status for diverse purposes and suggests accurate interpolation method for visualizing $CO_2$ emission as the basic data for regulating and managing $CO_2$ emission by applying IDW, Spline, and Kringing method. Results showed that Gaussian Function application among the Kriging methods had the highest accuracy in its estimations, with 3.049 with RMSE standards. This could be used as the basic data when visualizing $CO_2$ emission status, which is a necessity for many local and federal governments that are to regulate and manage $CO_2$ emission. This study shows that the interpolation is very appropriative method in recognizing $CO_2$ emission characteristics for regional climate change measures.

• 한국산업보건학회지
• /
• 제28권3호
• /
• pp.267-272
• /
• 2018

Objective: This study aimed to test the accuracy of self-reported information used in indirect estimation of pesticide exposure. Methods: To do so, self-reported values on the duration of pesticide application per day were compared with observed values. The number of days of pesticide application per year as recorded in self-administered logs was compared with recalled values. Results: It was found that participants underestimated the duration and frequency of actual pesticide use. High correlations were found between self-reported values and observed values, as well as between recalled values and recorded values. Conclusions: The reason might be that farmers unconsciously under-recall the application of pesticide since many customers prefer eco-friendly agricultural products. Farmers thought the task of applying pesticides to be essential, and this may explain why the participants in this study tended to accurately recall their pesticide-related work.

• 대한원격탐사학회지
• /
• 제18권2호
• /
• pp.81-90
• /
• 2002

본 논문에서는 대전과 논산지역의 KOMPSAT-1 EOC입체 영상으로부터 DEM을 생성하고 정확도를 검증하였다. DEM생성 과정을 크게 카메라 모델링 단계와 영상 정합 단계로 구분하여 논의하였으며 카메라 모델링 기법은 Orun과 Natarajan이 제안한 모델(1994)과 Gupta와 Harteley(1997)가 제안한 DLT모델을 사용하였으며 두 모델링 기법을 EOC입체 영상에 적용하는 것이 가능한지 확인하였다. 영상정합 단계에서는SPOT용으로 개발된 알고리즘이 EOC입체 영상에 적용될 수 있는지를 검토 하였다. 그리고 각 단계마다 EOC영상에 적용했을 때의 결과를 SPOT영상을 적용했을 때의 결과와 비교하였다. 본 실험에서 KOMPSAT-1 EOC입체 영상에 대해 카메라 모델링 기법과 영상 정합을 수행하여 DEM을 생성한 결과 SPOT입체 영상에서 생성한 DEM 보다 성능이 우수한 DEM을 얻을 수 있었다.

• 대한원격탐사학회지
• /
• 제37권3호
• /
• pp.649-655
• /
• 2021

본 연구는 경주시를 대상으로 수행한 선행연구를 바탕으로 도출된 최적의 지진 취약성 평가 모델을 타 지역에 적용하여 그 성능을 교차 검증(cross-validation)하고자 한다. 테스트 지역은 2017 포항지진(Pohang Earthquake)이 발생한 포항시이며, 선행연구와 동일한 영향인자 및 피해현황 관련 데이터셋을 구축하였다. 검증 데이터 셋은 무작위로 추출해 구축하였으며, 경주시의 랜덤 포레스트(random forest, RF) 기반의 모델에 적용하여 예측 정확도를 도출하였다. 경주시의 모델(success) 및 예측(prediction) 정확도는 100%, 94.9%이며, 포항시 검증 데이터 셋을 적용해 예측 정확도를 확인한 결과 70.4%로 나타났다.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2006년도 Proceedings of ISRS 2006 PORSEC Volume II
• /
• pp.643-646
• /
• 2006

This paper addresses the approach to extract linear features from satellite imagery using an efficient segmentation method. The extraction of linear features from satellite images has been the main concern of many scientists. There is a need to develop a more capable and cost effective method for the Iranian map revision tasks. The conventional approaches for producing, maintaining, and updating GIS map are time consuming and costly process. Hence, this research is intended to investigate how to obtain linear features from SPOT satellite imagery. This was accomplished using a discontinuity-based segmentation technique that encompasses four stages: low level bottom-up, middle level bottom-up, edge thinning and accuracy assessment. The first step is geometric correction and noise removal using suitable operator. The second step includes choosing the appropriate edge detection method, finding its proper threshold and designing the built-up image. The next step is implementing edge thinning method using mathematical morphology technique. Lastly, the geometric accuracy assessment task for feature extraction as well as an assessment for the built-up result has been carried out. Overall, this approach has been applied successfully for linear feature extraction from SPOT image.