• Journal of the Korea Society of Computer and Information
• /
• v.26 no.1
• /
• pp.27-35
• /
• 2021

In this paper, we propose a new stacking ensemble framework for deep learning models which reflects the distribution of label embeddings. Our ensemble framework consists of two phases: training the baseline deep learning classifier, and training the sub-classifiers based on the clustering results of label embeddings. Our framework aims to divide a multi-class classification problem into small sub-problems based on the clustering results. The clustering is conducted on the label embeddings obtained from the weight of the last layer of the baseline classifier. After clustering, sub-classifiers are constructed to classify the sub-classes in each cluster. From the experimental results, we found that the label embeddings well reflect the relationships between classification labels, and our ensemble framework can improve the classification performance on a CIFAR 100 dataset.

• International journal of advanced smart convergence
• /
• v.11 no.2
• /
• pp.13-21
• /
• 2022

In this paper, we propose a learning based joint demosaicing and super-resolution framework which uses only the mosaiced color filter array(CFA) image as the input. As the proposed method works only on the mosaicied CFA image itself, there is no need for a large dataset. Based on our framework, we proposed two different structures, where the first structure uses one deep image prior network, while the second uses two. Experimental results show that even though we use only the CFA image as the training image, the proposed method can result in better visual quality than other bilinear interpolation combined demosaicing methods, and therefore, opens up a new research area for joint demosaicing and super-resolution on raw images.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.6
• /
• pp.2012-2030
• /
• 2021

Named Entity Recognition (NER) is used to identify entity nouns in the corpus such as Location, Person and Organization, etc. NER is also an important basic of research in various natural language fields. The processing of Chinese NER has some unique difficulties, for example, there is no obvious segmentation boundary between each Chinese character in a Chinese sentence. The Chinese NER task is often combined with Chinese word segmentation, and so on. In response to these problems, we summarize the recognition methods of Chinese NER. In this review, we first introduce the sequence labeling system and evaluation metrics of NER. Then, we divide Chinese NER methods into rule-based methods, statistics-based machine learning methods and deep learning-based methods. Subsequently, we analyze in detail the model framework based on deep learning and the typical Chinese NER methods. Finally, we put forward the current challenges and future research directions of Chinese NER technology.

• Journal of the Society of Naval Architects of Korea
• /
• v.56 no.6
• /
• pp.473-479
• /
• 2019

Welding is one of the most popular joining methods and most welding quality estimation methods are executed using joined material. This paper propose welding quality estimation methods using dynamic current, voltage and resistance which are obtained during welding in real time. There are many kinds of welding method. Among them, we focused on the projection welding and gathered dynamic characteristics from two different types of projection welding. For image learning, graphs are drawn using obtained current, voltage and resistance, and the graphs are converted to images. The images are labeled with two sub-categories - normal and defect. For deep learning of images obtained from welding, Convolutional Neural Network (CNN) is applied, and Tensorflow was used as a framework for deep learning. With two resistance welding test datasets, we conclude that the Convolutional Neural Network helps in predicting the welding quality.

• Journal of Software Assessment and Valuation
• /
• v.15 no.1
• /
• pp.63-70
• /
• 2019

These days, as Interest in Image recognition with deep learning is increasing, there has been a lot of research in image recognition using deep learning. In this study, we propose a system for classifying rocks through rock images of 18 types of rock(6 types of igneous, 6 types of metamorphic, 6 types of sedimentary rock) which are addressed in the high school curriculum, using CNN model based on Tensorflow, deep learning open source framework. As a result, we developed a classifier to distinguish rocks by learning the images of rocks and confirmed the classification performance of rock classifier. Finally, through the mobile application implemented, students can use the application as a learning tool in classroom or on-site experience.

• Korean Journal of Remote Sensing
• /
• v.33 no.4
• /
• pp.423-436
• /
• 2017

This paper presents a deep learning-based road segmentation framework from very high-resolution orthophotos. The proposed method uses Deep Convolutional Autoencoders for end-to-end mapping of orthophotos to road segmentations. In addition, a set of post-processing steps were applied to make the model outputs GIS-ready data that could be useful for various applications. The optimization of the model's parameters is explained which was conducted via grid search method. The model was trained and implemented in Keras, a high-level deep learning framework run on top of Tensorflow. The results show that the proposed model with the best-obtained hyperparameters could segment road objects from orthophotos at an average accuracy of 88.5%. The results of optimization revealed that the best optimization algorithm and activation function for the studied task are Stochastic Gradient Descent (SGD) and Exponential Linear Unit (ELU), respectively. In addition, the best numbers of convolutional filters were found to be 8 for the first and second layers and 128 for the third and fourth layers of the proposed network architecture. Moreover, the analysis on the time complexity of the model showed that the model could be trained in 4 hours and 50 minutes on 1024 high-resolution images of size $106{\times}106pixels$, and segment road objects from similar size and resolution images in around 14 minutes. The results show that the deep learning models such as Convolutional Autoencoders could be a best alternative to traditional machine learning models for road segmentation from aerial photographs.

• Journal of Broadcast Engineering
• /
• v.27 no.3
• /
• pp.341-350
• /
• 2022

In this paper, we propose a deep learning based on-device augmented reality (AR) system in which multiple input images are used to implement the correct occlusion in a real environment. The proposed system is composed of three technical steps; camera pose estimation, depth estimation, and object augmentation. Each step employs various mobile frameworks to optimize the processing on the on-device environment. Firstly, in the camera pose estimation stage, the massive computation involved in feature extraction is parallelized using OpenCL which is the GPU parallelization framework. Next, in depth estimation, monocular and multiple image-based depth image inference is accelerated using the mobile deep learning framework, i.e. TensorFlow Lite. Finally, object augmentation and occlusion handling are performed on the OpenGL ES mobile graphics framework. The proposed augmented reality system is implemented as an application in the Android environment. We evaluate the performance of the proposed system in terms of augmentation accuracy and the processing time in the mobile as well as PC environments.

• Journal of KIBIM
• /
• v.10 no.4
• /
• pp.60-69
• /
• 2020

Allowing on-site workers to access information of the construction process can enable task control, data integration, material and resource control. However, in the current practice of the construction industry, the existing methods and scope is quite limited, leading to inefficient management during the construction process. In this research, by adopting cutting edge technologies such as Augmented Reality(AR), digital twins, deep learning and computer vision with wearable AR devices, the authors proposed an AR visualization framework made of virtual components to help on-site workers to obtain information of the construction process with ease of use. Also, this paper investigates wearable AR devices and object detection algorithms, which are critical factors in the proposed framework, to test their suitability.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2019.05a
• /
• pp.385-388
• /
• 2019

In recent years, China has faced unprecedented intelligent reforms. Artificial intelligence has become a hot topic in society. The deep learning framework is the core of artificial intelligence industrialization, and it has also attracted the attention of all parties. Among them, deep learning has been applied in the fields of computer vision, speech recognition, and language technology processing. This paper will introduce China's development status and future challenges in technology, talent, and market applications.

• International Journal of Computer Science & Network Security
• /
• v.22 no.11
• /
• pp.265-271
• /
• 2022

Social media is a form of communication based on the internet to share information through content and images. Their choice of profile images and type of image they post can be closely connected to their personality. The user posted images are designated as personality traits. The objective of this study is to predict five factor model personality dimensions from profile images by using deep learning and neural networks. Developed a deep learning framework-based neural network for personality prediction. The personality types of the Big Five Factor model can be quantified from user profile images. To measure the effectiveness, proposed two models using convolution Neural Networks to classify each personality of the user. Done performance analysis among two different models for efficiently predict personality traits from profile image. It was found that VGG-69 CNN models are best performing models for producing the classification accuracy of 91% to predict user personality traits.