• Journal of Korea Multimedia Society
• /
• v.21 no.12
• /
• pp.1387-1395
• /
• 2018

Core algorithm of deep learning Convolutional Neural Network(CNN) shows better performance than other machine learning algorithms. However, if there is not sufficient data, CNN can not achieve satisfactory performance even if the classifier is excellent. In this situation, it has been proven that the use of transfer learning can have a great effect. In this paper, we apply two transition learning methods(freezing, retraining) to three CNN models(ResNet-50, Inception-V3, DenseNet-121) and compare and analyze how the classification performance of CNN changes according to the methods. As a result of statistical significance test using various evaluation indicators, ResNet-50, Inception-V3, and DenseNet-121 differed by 1.18 times, 1.09 times, and 1.17 times, respectively. Based on this, we concluded that the retraining method may be more effective than the freezing method in case of transition learning in image classification problem.

• ETRI Journal
• /
• v.45 no.2
• /
• pp.318-328
• /
• 2023

Recently, embedded systems, such as mobile platforms, have multiple processing units that can operate in parallel, such as centralized processing units (CPUs) and neural processing units (NPUs). We can use deep-learning compilers to generate machine code optimized for these embedded systems from a deep neural network (DNN). However, the deep-learning compilers proposed so far generate codes that sequentially execute DNN operators on a single processing unit or parallel codes for graphic processing units (GPUs). In this study, we propose PartitionTuner, an operator scheduler for deep-learning compilers that supports multiple heterogeneous PUs including CPUs and NPUs. PartitionTuner can generate an operator-scheduling plan that uses all available PUs simultaneously to minimize overall DNN inference time. Operator scheduling is based on the analysis of DNN architecture and the performance profiles of individual and group operators measured on heterogeneous processing units. By the experiments for seven DNNs, PartitionTuner generates scheduling plans that perform 5.03% better than a static type-based operator-scheduling technique for SqueezeNet. In addition, PartitionTuner outperforms recent profiling-based operator-scheduling techniques for ResNet50, ResNet18, and SqueezeNet by 7.18%, 5.36%, and 2.73%, respectively.

• Journal of the Architectural Institute of Korea Structure & Construction
• /
• v.36 no.3
• /
• pp.113-120
• /
• 2020

The purpose of this study is to compare the models of Deep Learning-based Convolution Neural Network(CNN) for concrete crack detection. The comparison models are AlexNet, GoogLeNet, VGG16, VGG19, ResNet-18, ResNet-50, ResNet-101, and SqueezeNet which won ImageNet Large Scale Visual Recognition Challenge(ILSVRC). To train, validate and test these models, we constructed 3000 training data and 12000 validation data with 256×256 pixel resolution consisting of cracked and non-cracked images, and constructed 5 test data with 4160×3120 pixel resolution consisting of concrete images with crack. In order to increase the efficiency of the training, transfer learning was performed by taking the weight from the pre-trained network supported by MATLAB. From the trained network, the validation data is classified into crack image and non-crack image, yielding True Positive (TP), True Negative (TN), False Positive (FP), False Negative (FN), and 6 performance indicators, False Negative Rate (FNR), False Positive Rate (FPR), Error Rate, Recall, Precision, Accuracy were calculated. The test image was scanned twice with a sliding window of 256×256 pixel resolution to classify the cracks, resulting in a crack map. From the comparison of the performance indicators and the crack map, it was concluded that VGG16 and VGG19 were the most suitable for detecting concrete cracks.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.59-64
• /
• 2024

Recently, research on the convergence of drones and artificial intelligence technologies have been conducted in various industrial fields. In this paper, we propose an illegal parking vehicle recognition model using deep learning-based object recognition and classification algorithms. The model of object recognition and classification consist of YOLOv8 and ResNet18, respectively. The proposed model was trained using image data collected in general road environment, and the trained model showed high accuracy in determining illegal parking. From simulation results, it was confirmed that the proposed model has generalization performance to identify illegal parking vehicles from various images.

• Imaging Science in Dentistry
• /
• v.54 no.3
• /
• pp.257-263
• /
• 2024

Purpose: The purpose of this study was to classify mandibular molar furcation involvement (FI) in periapical radiographs using a deep learning algorithm. Materials and Methods: Full mouth series taken at East Carolina University School of Dental Medicine from 2011-2023 were screened. Diagnostic-quality mandibular premolar and molar periapical radiographs with healthy or FI mandibular molars were included. The radiographs were cropped into individual molar images, annotated as "healthy" or "FI," and divided into training, validation, and testing datasets. The images were preprocessed by PyTorch transformations. ResNet-18, a convolutional neural network model, was refined using the PyTorch deep learning framework for the specific imaging classification task. CrossEntropyLoss and the AdamW optimizer were employed for loss function training and optimizing the learning rate, respectively. The images were loaded by PyTorch DataLoader for efficiency. The performance of ResNet-18 algorithm was evaluated with multiple metrics, including training and validation losses, confusion matrix, accuracy, sensitivity, specificity, the receiver operating characteristic (ROC) curve, and the area under the ROC curve. Results: After adequate training, ResNet-18 classified healthy vs. FI molars in the testing set with an accuracy of 96.47%, indicating its suitability for image classification. Conclusion: The deep learning algorithm developed in this study was shown to be promising for classifying mandibular molar FI. It could serve as a valuable supplemental tool for detecting and managing periodontal diseases.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.8
• /
• pp.2103-2118
• /
• 2024

Chinese herbal slices (CHS) automated recognition based on computer vision plays a critical role in the practical application of intelligent Chinese medicine. Due to the complexity and similarity of herbal images, identifying Chinese herbal slices is still a challenging task. Especially, easily-confused CHS have higher inter-class and intra-class complexity and similarity issues, the existing deep learning models are less adaptable to identify them efficiently. To comprehensively address these problems, a novel tiny easily-confused CHS dataset has been built firstly, which includes six pairs of twelve categories with about 2395 samples. Furthermore, we propose a ResNeSt-CHS model that combines multilevel perception fusion (MPF) and perceptive sparse fusion (PSF) blocks for efficiently recognizing easilyconfused CHS images. To verify the superiority of the ResNeSt-CHS and the effectiveness of our dataset, experiments have been employed, validating that the ResNeSt-CHS is optimal for easily-confused CHS recognition, with 2.1% improvement of the original ResNeSt model. Additionally, the results indicate that ResNeSt-CHS is applied on a relatively small-scale dataset yet high accuracy. This model has obtained state-of-the-art easily-confused CHS classification performance, with accuracy of 90.8%, far beyond other models (EfficientNet, Transformer, and ResNeSt, etc) in terms of evaluation criteria.

• Convergence Security Journal
• /
• v.23 no.5
• /
• pp.57-64
• /
• 2023

In this study, we analyzed how to steal the target model without training data. Input data is generated using the generative model, and a similar model is created by defining a loss function so that the predicted values of the target model and the similar model are close to each other. At this time, the target model has a process of learning so that the similar model is similar to it by gradient descent using the logit (logic) value of each class for the input data. The tensorflow machine learning library was used as an experimental environment, and CIFAR10 and SVHN were used as datasets. A similar model was created using the ResNet model as a target model. As a result of the experiment, it was found that the model stealing method generated a similar model with an accuracy of 86.18% for CIFAR10 and 96.02% for SVHN, producing similar predicted values to the target model. In addition, considerations on the model stealing method, military use, and limitations were also analyzed.

• Journal of Information Processing Systems
• /
• v.18 no.2
• /
• pp.197-207
• /
• 2022

The applications and user interfaces (UIs) of smart mobile devices are constantly diversifying. For example, deep learning can be an innovative solution to classify widgets in screen images for increasing convenience. To this end, the present research leverages captured images and the ReDraw dataset to write deep learning datasets for image classification purposes. First, as the validation for datasets using ResNet50 and EfficientNet, the experiments show that the dataset composed in this study is helpful for classification according to a widget's functionality. An implementation for widget detection and classification on RetinaNet and EfficientNet is then executed. Finally, the research suggests the Widg-C and Widg-D datasets-a deep learning dataset for identifying the widgets of smart devices-and implementing them for use with representative convolutional neural network models.

• Korean journal of applied entomology
• /
• v.46 no.1 s.145
• /
• pp.153-158
• /
• 2007

This study was conducted to develope the observing methods for density of stink bugs in soybean reproductive stage. The adults and nymphs of bean bug, Riptortus clavatus, red-banded shield bug, Piezodous hybneri, green stink bug, Nezara antennata, Sole bug, Dolycoris baccarum, and brown marmorated stink bug, Halyomorpha halys were observed by three observing methods such as beating, sweeping net, and visual counting methods in the full bloom (R2), full pod (R4) and beginning maturity (R7) of soybean. As a result, total number of stink bugs observed was the highest with 5,214.2 by beating method, and then was 2,581.8 by visual counting method, and was the lowest with 103.1 by sweeping net method. Total number of stink bugs observed by the beating and visual counting methods was P. hybneri, followed by N. antennata, H. halys, R. clavatus and D. baccarum with clear difference in observed number of each stink bugs while total number of stink bugs observed by sweeping net method was very low in the range of 18 to 23. Accordingly, the observed density of stink bugs exception of R. clavatus adult by beating method was generally high. However, the number of R. clavatus adult was more observed by flushing method than that by beating method from the beginning bloom (R1) to full maturity (R8), and was more observed at morning time than that at afternoon time. Therefore, two observation methods that flushing method for R. clavatus and beating method for the other stink bugs were recommended for the occurring density of stink bugs in soybean because both bean bug and pentatomidae stink bugs have distinct behavior characteristics such as flying and dropping.

• Structural Engineering and Mechanics
• /
• v.91 no.2
• /
• pp.163-175
• /
• 2024

The objective of this research is to improve public safety in civil engineering by recognizing fractures in concrete structures quickly and correctly. The study offers a new crack detection method based on advanced image processing and machine learning techniques, specifically transfer learning with convolutional neural networks (CNNs). Four pre-trained models (VGG16, AlexNet, ResNet18, and DenseNet161) were fine-tuned to detect fractures in concrete surfaces. These models constantly produced accuracy rates greater than 80%, showing their ability to automate fracture identification and potentially reduce structural failure costs. Furthermore, the study expands its scope beyond crack detection to identify concrete health, using a dataset with a wide range of surface defects and anomalies including cracks. Notably, using VGG16, which was chosen as the most effective network architecture from the first phase, the study achieves excellent accuracy in classifying concrete health, demonstrating the model's satisfactorily performance even in more complex scenarios.