This study aims to develop a deep learning model to monitor rice serving amounts in institutional foodservice, enhancing personalized nutrition management. The goal is to identify the best convolutional neural network (CNN) for detecting rice quantities on serving trays, addressing balanced dietary intake challenges. Both a vanilla CNN and 12 pre-trained CNNs were tested, using features extracted from images of varying rice quantities on white trays. Configurations included optimizers, image generation, dropout, feature extraction, and fine-tuning, with top-1 validation accuracy as the evaluation metric. The vanilla CNN achieved 60% top-1 validation accuracy, while pre-trained CNNs significantly improved performance, reaching up to 90% accuracy. MobileNetV2, suitable for mobile devices, achieved a minimum 76% accuracy. These results suggest the model can effectively monitor rice servings, with potential for improvement through ongoing data collection and training. This development represents a significant advancement in personalized nutrition management, with high validation accuracy indicating its potential utility in dietary management. Continuous improvement based on expanding datasets promises enhanced precision and reliability, contributing to better health outcomes.

Liquefied hydrogen is attracting attention as an energy source of the future due to its hydrogen storage rate and low risk. However, the disadvantage is that the unit price is high due to technical difficulties in production, transportation, and storage. This study was conducted to improve the design accuracy and development period of needle valves, which are important parts with a wide technical application range among liquefied hydrogen equipment. Since the needle valve must discharge an appropriate flow rate of the liquefied fluid, it is important to determine the needle valve design parameters suitable for the target flow rate. Computational Fluid Dynamics and Artificial Neural Network technology used to determine the design variables of fluid flow were applied to improve the setting and analysis time of the parameter. In addition, procedures and methods for applying the design parameter of needle valves to Convolutional Neural Networks were presented. The procedure and appropriate conditions for selecting parameters and functional conditions of the Convolutional Neural Network were presented, and the accuracy of predicting the flow coefficient according to the design parameter was secured 95%. It is judged that this method can be applied to other structures and machines.

Determining the size or area of a plant's leaves is an important factor in predicting plant growth and improving the productivity of indoor farms. In this study, we developed a convolutional neural network (CNN)-based model to accurately predict the length and width of lettuce leaves using photographs of the leaves. A callback function was applied to overcome data limitations and overfitting problems, and K-fold cross-validation was used to improve the generalization ability of the model. In addition, ImageDataGenerator function was used to increase the diversity of training data through data augmentation. To compare model performance, we evaluated pre-trained models such as VGG16, Resnet152, and NASNetMobile. As a result, NASNetMobile showed the highest performance, especially in width prediction, with an R_squared value of 0.9436, and RMSE of 0.5659. In length prediction, the R_squared value was 0.9537, and RMSE of 0.8713. The optimized model adopted the NASNetMobile architecture, the RMSprop optimization tool, the MSE loss functions, and the ELU activation functions. The training time of the model averaged 73 minutes per Epoch, and it took the model an average of 0.29 seconds to process a single lettuce leaf photo. In this study, we developed a CNN-based model to predict the leaf length and leaf width of plants in indoor farms, which is expected to enable rapid and accurate assessment of plant growth status by simply taking images. It is also expected to contribute to increasing the productivity and resource efficiency of farms by taking appropriate agricultural measures such as adjusting nutrient solution in real time.

In factory automation, efforts are being made to increase productivity while maintaining high-quality products. In this study, a CNN network structure was designed to quickly and accurately recognize a cigarette located in the opposite direction or a cigarette with a loose end in an automated facility rotating at high speed for cigarette production. Tobacco inspection requires a simple network structure and fast processing time and performance. The proposed network has an excellent accuracy of 96.33% and a short processing time of 0.527 msec, showing excellent performance in learning time and performance compared to other CNN networks, confirming its practicality. In addition, it was confirmed that efficient learning is possible by increasing a small number of image data through a rotation conversion method.

본 논문에서는 볼트로 체결된 구조체에 대하여 초기 볼트풀림 상태에서의 볼트 체결력 예측 합성곱 신경망 훈련 방법을 제시한다. 8개의 볼트의 체결력이 변경된 상태에서 계산한 주파수응답들을 완전 체결된 상태의 초기 모델과의 크기 및 모양 유사성을 표현하는 유사성 지도로 생성한다. 주파수응답 데이터들의 생성에는 크리로프 부공간법 기반의 모델차수축소법을 적용하여 효율적인 방법으 로 수행할 수 있도록 한다. 합성곱 신경망 모델은 회귀 출력 계층을 사용하여 볼트의 체결력을 예측하도록 하였으며, 훈련 데이터의 개 수와 합성곱 신경망 계층의 개수를 다르게 준비하여 훈련시킨 네트워크들을 비교하여 그 성능을 평가하였다. 주파수응답에서 파생되 는 유사성 지도를 입력 데이터로 사용하여 초기 볼트풀림 영역에서 볼트 체결력의 진단 가능성과 유효성을 제시하였다.

The purpose of this study was to verify the sensitive areas when the AI determines osteoporosis for the entire area of the panoramic radiograph. Panoramic radiographs of a total of 1,156 female patients(average age of 49.0±24.0 years) were used for this study. The panoramic radiographs were diagnosed as osteoporosis and the normal by Oral and Maxillofacial Radiology specialists. The VGG16 deep learning convolutional neural network(CNN) model was used to determine osteoporosis and the normal from testing 72 osteoporosis(average age of 73.7±8.0 years) and 93 normal(average age of 26.4±5.1 years). VGG16 conducted a gradient-weighted class activation mapping(Grad-CAM) visualization to indicate sensitive areas when determining osteoporosis. The accuracy of CNN in determining osteoporosis was 100%. Heatmap image from 72 panoamic radiographs of osteoporosis revealed that CNN was sensitive to the cervical vertebral in 70.8%(51/72), the cortical bone of the lower mandible in 72.2%(52/72), the cranial base area in 30.6%(22/72), the cancellous bone of the mandible in 33.3%(24/72), the cancellous bone of the maxilla in 20.8%(15/72), the zygoma in 8.3%(6/72), and the dental area in 5.6%(4/72). Consideration: it was found that the cervical vertebral area and the cortical bone of the lower mandible were sensitive areas when CNN determines osteoporosis in the entire area of panoramic radiographs.

본 연구는 무대재배 복숭아 ‘미황’을 대상으로 성숙기간 중 RGB 영상을 취득한 후 다양한 품질 지표를 측정하고 이를 딥 러닝 기술에 적용하여 복숭아 과실 숙도 분류의 가능성을 탐 색하고자 실시하였다. 취득 영상 730개의 데이터를 training 과 validation에 사용하였고, 170개는 최종 테스트 이미지로 사용하였다. 본 연구에서는 딥러닝을 활용한 성숙도 자동 분 류를 위하여 조사된 품질 지표 중 경도, Hue 값, a*값을 최종 선 발하여 이미지를 수동으로 미성숙(immature), 성숙(mature), 과숙(over mature)으로 분류하였다. 이미지 자동 분류는 CNN (Convolutional Neural Networks, 컨볼루션 신경망) 모델 중 에서 이미지 분류 및 탐지에서 우수한 성능을 보이고 있는 VGG16, GoogLeNet의 InceptionV3 두 종류의 모델을 사용 하여 복숭아 품질 지표 값의 분류 이미지별 성능을 측정하였 다. 딥러닝을 통한 성숙도 이미지 분석 결과, VGG16과 InceptionV3 모델에서 Hue_left 특성이 각각 87.1%, 83.6% 의 성능(F1 기준)을 나타냈고, 그에 비해 Firmness 특성이 각각 72.2%, 76.9%를 나타냈고, Loss율이 각각 54.3%, 62.1% 로 Firmness를 기준으로 한 성숙도 분류는 적용성이 낮음을 확인하였다. 추후에 더 많은 종류의 이미지와 다양한 품질 지 표를 가지고 학습이 진행된다면 이전 연구보다 향상된 정확도 와 세밀한 성숙도 판별이 가능할 것으로 판단되었다.

Visual inspection methods have limitations, such as reflecting the subjective opinions of workers. Moreover, additional equipment is required when inspecting the high-rise buildings because the height is limited during the inspection. Various methods have been studied to detect concrete cracks due to the disadvantage of existing visual inspection. In this study, a crack detection technology was proposed, and the technology was objectively and accurately through AI. In this study, an efficient method was proposed that automatically detects concrete cracks by using a Convolutional Neural Network(CNN) with the Orthomosaic image, modeled with the help of UAV. The concrete cracks were predicted by three different CNN models: AlexNet, ResNet50, and ResNeXt. The models were verified by accuracy, recall, and F1 Score. The ResNeXt model had the high performance among the three models. Also, this study confirmed the reliability of the model designed by applying it to the experiment.

In this study, the multi-lane detection problem is expressed as a CNN-based regression problem, and the lane boundary coordinates are selected as outputs. In addition, we described lanes as fifth-order polynomials and distinguished the ego lane and the side lanes so that we could make the prediction lanes accurately. By eliminating the network branch arrangement and the lane boundary coordinate vector outside the image proposed by Chougule’s method, it was possible to eradicate meaningless data learning in CNN and increase the fast training and performance speed. And we confirmed that the average prediction error was small in the performance evaluation even though the proposed method compared with Chougule’s method under harsher conditions. In addition, even in a specific image with many errors, the predicted lanes did not deviate significantly, meaningful results were derived, and we confirmed robust performance.

Deep convolutional network is a deep learning approach to optimize image recognition. This study aimed to apply DCNN to the reading of mandibular cortical thinning in digital panoramic radiographs. Digital panoramic radiographs of 1,268 female dental patients (age 45.2 ± 21.1yrs) were used in the reading of the mandibular cortical bone by two maxillofacial radiologists. Among the subjects, 535 normal subject’s panoramic radiographs (age 28.6 ±7.4 yrs) and 533 those of osteoporosis pationts (age 72.1 ± 8.7 yrs) with mandibular cortical thinning were used for training DCNN. In the testing of mandibular cortical thinning, 100 panoramic radiographs of normal subjects (age 26.6 ± 4.5 yrs) and 100 mandibular cortical thinning (age 72.5 ± 7.2 yrs) were used. The sensitive area of DCNN to mandibular cortical thinning was investigated by occluding analysis. The readings of DCNN were compared by two maxillofacial radiologists. DCNN showed 97.5% accuracy, 96% sensitivity, and 99% specificity in reading mandibular cortical thinning. DCNN was sensitively responded on the cancellous and cortical bone of the mandibular inferior area. DCNN was effective in diagnosing mandibular cortical thinning.

Gravity Recovery and Climate Experiment (GRACE) gravimeter satellites observed the Earth gravity field with unprecedented accuracy since 2002. After the termination of GRACE mission, GRACE Follow-on (GFO) satellites successively observe global gravity field, but there is missing period between GRACE and GFO about one year. Many previous studies estimated terrestrial water storage (TWS) changes using hydrological models, vertical displacements from global navigation satellite system observations, altimetry, and satellite laser ranging for a continuity of GRACE and GFO data. Recently, in order to predict TWS changes, various machine learning methods are developed such as artificial neural network and multi-linear regression. Previous studies used hydrological and climate data simultaneously as input data of the learning process. Further, they excluded linear trends in input data and GRACE/GFO data because the trend components obtained from GRACE/GFO data were assumed to be the same for other periods. However, hydrological models include high uncertainties, and observational period of GRACE/GFO is not long enough to estimate reliable TWS trends. In this study, we used convolutional neural networks (CNN) method incorporating only climate data set (temperature, evaporation, and precipitation) to predict TWS variations in the missing period of GRACE/GFO. We also make CNN model learn the linear trend of GRACE/GFO data. In most river basins considered in this study, our CNN model successfully predicts seasonal and long-term variations of TWS change.

자기공명영상은 고해상도의 연부조직에 대한 영상정보를 제공하며, 뇌종양 등 연부조직 진단에 활용된다. 본 연구는 합성곱신경망 인공지능을 통해 뇌종양 자기공명영상 분류성능을 확인해 보고자 한다. 4개 종류로 구분된 3264 장의 MRI 데이터 세트(data set)를 이용하였으며, 인공지능 학습을 위해 훈련용 데이터와 시험용 데이터를 9 : 1, 훈련용 데이터의 10%를 검증용 데이터로 구분하였다. 합성곱신경망은 기본 CNN과 VGG16으로 구성하였으며, 학습 평가는 정확도와 손실율로 확인하였으며, 생성된 모델을 통해 분류성능 정확도를 확인하였다. 실험 결과 과적합은 없었으며, 분류성능은 기본 CNN과 VGG16 각각 67%와 80%의 분류성능을 보였다. 도출된 뇌종양 자기공명영상 분류 결과를 통해 자기공명영상과 인공지능 접목에 관한 기초 자료로 사용될 수 있을 것이라 사료된다.

The important thing in the field of deep learning is to find out the appropriate hyper-parameter for image classification. In this study, the main objective is to investigate the performance of various hyper-parameters in a convolutional neural network model based on the image classification problem. The dataset was obtained from the Kaggle dataset. The experiment was conducted through different hyper-parameters. For this proposal, Stochastic Gradient Descent without momentum (SGD), Adaptive Moment Estimation (Adam), Adagrad, Adamax optimizer, and the number of batch sizes (16, 32, 64, 120), and the number of epochs (50, 100, 150) were considered as hyper-parameters to determine the losses and accuracy of a model. In addition, Binary Cross-entropy Loss Function (BCLF) was used for evaluating the performance of a model. In this study, the VGG16 convolutional neural network was used for image classification. Empirical results demonstrated that a model had minimum losses obtain by Adagrad optimizer in the case of 16 batch sizes and 50 epochs. In addition, the SGD with a 32 batch sizes and 150 epochs and the Adam with a 64 batch sizes and 50 epochs had the best performance based on the loss value during the training process. Interestingly, the accuracy was higher while performing the Adagrad and Adamax optimizer with a 120 batch sizes and 150 epochs. In this study, the Adagrad optimizer with a 120 batch sizes and 150 epochs performed slightly better among those optimizers. In addition, an increasing number of epochs can improve the performance of accuracy. It can help to create a broader scope for further experiments on several datasets to perceive the suitable hyper-parameters for the convolutional neural network. Dataset: https://www.kaggle.com/c/dogs-vs-cats/data

작물의 생체중을 추정하기 위해 다양한 연구가 시도되었지만, 이미지를 활용하여 생체중을 추정한 예는 없었다. 최근 합성곱 신경망을 사용한 이미지 처리 연구가 늘고 있으며, 합성곱 신경망은 미가공 데이터를 그대로 사용할 수 있다. 본 연구에서는 합성곱 신경망을 이용하여 미가공 데이터 상태인 특정 시점의 파프리카 이미지를 입력으로 작물의 생체중을 추정하도록 학습하였다. 실험은 파프리카(Capsicum annuum L.)를 재배하는 온실에서 수행하였다. 합성곱 신경망의 출력값인 생체중은 파괴조사를 통해 수집한 데이터를 기반으로 회귀 분석하였다. 학습된 합성곱 신경망의 결정 계수(R2)의 최고값은 0.95로 나타났다. 생체중 추정값은 실제 측정값과 매우 유사한 경향성을 보여주었다.

We apply a modified Convolutional Neural Network (CNN) model in conjunction with transfer learning to predict whether an active region (AR) would produce a ≥C-class or ≥M-class flare within the next 24 hours. We collect line-of-sight magnetogram samples of ARs provided by the SHARP from May 2010 to September 2018, which is a new data product from the HMI onboard the SDO. Based on these AR samples, we adopt the approach of shuffle-and-split cross-validation (CV) to build a database that includes 10 separate data sets. Each of the 10 data sets is segregated by NOAA AR number into a training and a testing data set. After training, validating, and testing our model, we compare the results with previous studies using predictive performance metrics, with a focus on the true skill statistic (TSS). The main results from this study are summarized as follows. First, to the best of our knowledge, this is the first time that the CNN model with transfer learning is used in solar physics to make binary class predictions for both ≥C-class and ≥M-class flares, without manually engineered features extracted from the observational data. Second, our model achieves relatively high scores of TSS = 0.640±0.075 and TSS = 0.526±0.052 for ≥M-class prediction and ≥C-class prediction, respectively, which is comparable to that of previous models. Third, our model also obtains quite good scores in five other metrics for both ≥C-class and ≥M-class flare prediction. Our results demonstrate that our modified CNN model with transfer learning is an effective method for flare forecasting with reasonable prediction performance.

This study was conducted as part of a series of studies to introduce the Convolutional Neural Network(CNN) into the diagnostic field of osteoporosis. The purpose of this study was to compare the results when testing Digital Radiography(DR) and Computed Radiography(CR) panoramic radiographs by CNN that were trained by DR panoramic radiographs. The digital panoramic radiographs of females who visited for the purpose of diagnosis and treatment at Chonnam National University Dental Hospital were taken. Two Oral and Maxillofacial Radiologists were selected for the study to compare the panoramic radiographs with normal and osteoporosis images. Among them, 1068 panoramic radiographs of females{Mean [± standard deviation] age: 49.19 ± 21.91 years} obtained by DR method were used for training of CNN. 200 panoramic radiographs of females{Mean [± standard deviation] age: 63.95 ± 6.45 years} obtained by DR method and 202 panoramic radiographs of females{Mean [± standard deviation] age: 62.00 ± 6.86 years} obtained by CR method were used for testing of CNN. When the DR panoramic radiographs were tested, the Accuracy was 92.5%. When the CR panoramic radiographs were tested, the Accuracy was 76.2%. It can be seen that the CNN trained by DR panoramic radiographs is suitable to be tested with the same DR panoramic radiographs.

화재의 초기 검출은 인명과 재화의 손실을 최소화하기 위한 중요한 요소이다. 불꽃과 연기를 신속하면서 동시에 검출해야 하며 이를 위해 영상 기반의 화재 검출에 관한 연구가 다양하게 진행되고 있다. 기존의 화재 검출은 불꽃과 연기의 특징을 추출하기 위해 여러 알고리즘을 거쳐서 화재의 검출 유무를 판단하므로 연산량이 많이 소모되었으나, 딥러닝 알고리즘인 합성곱 신경망을 이용 하면 별도의 과정이 생략되므로 신속하게 검출할 수 있다. 본 논문에서는 선박 기관실에서 화재 영상을 녹화한 데이터로 실험을 수행 하였다. 불꽃과 연기의 특징을 외각 상자로 추출한 후 합성곱 신경망 중 하나인 욜로(YOLO)를 이용하여 학습하고 결과를 테스트하였 다. 실험 결과를 검출률, 오검출률, 정확도로 평가하였으며 불꽃은 0.994, 0.011, 0.998, 연기는 0.978, 0.021, 0.978을 나타내었고, 연산시간 은 0.009s를 소모됨을 확인하였다.