• Journal of IKEEE
• /
• v.22 no.4
• /
• pp.1131-1139
• /
• 2018

This paper proposes a method of recognition abandoned objects using convolutional neural networks. The method first detects an area for an abandoned object in image and, if there is a detected area, applies convolutional neural networks to that area to recognize which object is represented. Experiments were conducted through an application system that detects illegal trash dumping. The experiments result showed the area of abandoned object was detected efficiently. The detected areas enter the input of convolutional neural networks and are classified into whether it is a trash or not. To do this, I trained convolutional neural networks with my own trash dataset and open database. As a training result, I achieved high accuracy for the test set not included in the training set.

• The Journal of Korean Institute of Next Generation Computing
• /
• v.15 no.3
• /
• pp.16-24
• /
• 2019

Recently, research on emotion recognition based on EEG has attracted great interest from human-robot interaction field. In this paper, we propose a method of labeling using image-based EEG topography instead of evaluating emotions through self-assessment and annotation labeling methods used in MAHNOB HCI. The proposed method evaluates the emotion by machine learning model that learned EEG signal transformed into topographical image. In the experiments using MAHNOB-HCI database, we compared the performance of training EEG topography labeling models of SVM and kNN. The accuracy of the proposed method was 54.2% in SVM and 57.7% in kNN.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.05a
• /
• pp.359-362
• /
• 2016

Face is an information unique to each person such as Iris, fingerprints, etc,. Research on face recognition are in progress continuously from the past to the present. Through these research, various face recognition methods have appeared. Among these methods, there are face recognition algorithms using the face data composed in stereo. In this paper, Convolutional Neural Network with Stereo Images as input was used for face recognition. This method showed better performance than the result of stereo face recognition using PCA that is used frequently in face recognition.

• KIPS Transactions on Computer and Communication Systems
• /
• v.11 no.7
• /
• pp.225-232
• /
• 2022

Recent development in information and communication technology has been beneficial to many, but at the same time, malicious attack attempts are also increasing through vulnerabilities in new programs. Among malicious attacks, malware operate in various ways and is distributed to people in new ways every time, and to solve this malware, it is necessary to quickly analyze and provide defense techniques. If new malware can be classified into the same type of malware, malware has similar behavioral characteristics, so they can provide defense techniques for new malware using analyzed malware. Therefore, there is a need for a solution to this because the method of accurately and quickly classifying malware and the number of data may not be uniform for each family of analyzed malware. This paper proposes a system that combines image preprocessing and ensemble techniques to increase accuracy in imbalanced data.

• Journal of IKEEE
• /
• v.27 no.3
• /
• pp.345-349
• /
• 2023

In this paper, we propose a study on the development of deep learning structure for defective pixel detection of next-generation smart LED display board using imaging device. In this research, a technique utilizing imaging devices and deep learning is introduced to automatically detect defects in outdoor LED billboards. Through this approach, the effective management of LED billboards and the resolution of various errors and issues are aimed. The research process consists of three stages. Firstly, the planarized image data of the billboard is processed through calibration to completely remove the background and undergo necessary preprocessing to generate a training dataset. Secondly, the generated dataset is employed to train an object recognition network. This network is composed of a Backbone and a Head. The Backbone employs CSP-Darknet to extract feature maps, while the Head utilizes extracted feature maps as the basis for object detection. Throughout this process, the network is adjusted to align the Confidence score and Intersection over Union (IoU) error, sustaining continuous learning. In the third stage, the created model is employed to automatically detect defective pixels on actual outdoor LED billboards. The proposed method, applied in this paper, yielded results from accredited measurement experiments that achieved 100% detection of defective pixels on real LED billboards. This confirms the improved efficiency in managing and maintaining LED billboards. Such research findings are anticipated to bring about a revolutionary advancement in the management of LED billboards.

• Journal of Internet Computing and Services
• /
• v.21 no.4
• /
• pp.9-16
• /
• 2020

In the Image processing and computer vision, the problem of reconstructing from one image to another or generating a new image has been steadily drawing attention as hardware advances. However, the problem of computer-generated images also continues to emerge when viewed with human eyes because it is not natural. Due to the recent active research in deep learning, image generating and improvement problem using it are also actively being studied, and among them, the network called Generative Adversarial Network(GAN) is doing well in the image generating. Various models of GAN have been presented since the proposed GAN, allowing for the generation of more natural images compared to the results of research in the image generating. Among them, pix2pix is a conditional GAN model, which is a general-purpose network that shows good performance in various datasets. pix2pix is based on U-Net, but there are many networks that show better performance among U-Net based networks. Therefore, in this study, images are generated by applying various networks to U-Net of pix2pix, and the results are compared and evaluated. The images generated through each network confirm that the pix2pix model with Attention, R2, and Attention-R2 networks shows better performance than the existing pix2pix model using U-Net, and check the limitations of the most powerful network. It is suggested as a future study.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.1
• /
• pp.29-34
• /
• 2021

The real-time video education is used as an effective method of operating classes that replaces face-to-face education of instructors and learners in remote areas. However, the existing video call and video conferences system is mainly used, and this is effective in linguistic education because it focuses on lecture through video, but it is not utilized in other education. In this paper, we propose a design model of real-time video system that can improve the effectiveness of science curriculum and mathematics education by providing the functions that can be utilized during class by improving limitations of image - oriented image education.

• Journal of the Semiconductor & Display Technology
• /
• v.20 no.3
• /
• pp.113-116
• /
• 2021

As artificial intelligence technologies, including deep learning, develop, these technologies are being introduced to code similarity analysis. In the traditional analysis method of calculating the graph edit distance (GED) after converting the source code into a control flow graph (CFG), there are studies that calculate the GED through a trained graph neural network (GNN) with the converted CFG, Methods for analyzing code similarity through CNN by imaging CFG are also being studied. In this paper, to determine which approach will be effective and efficient in researching code similarity analysis methods using artificial intelligence in the future, code similarity is measured through funcGNN, which measures code similarity using GNN, and Siamese Network, which is an image similarity analysis model. The accuracy was compared and analyzed. As a result of the analysis, the error rate (0.0458) of the Siamese network was bigger than that of the funcGNN (0.0362).

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.38 no.1
• /
• pp.69-83
• /
• 2020

Signboards are standardized according to national legislation for the safety of pedestrians and disaster prevention in urban areas. Signboards should be installed according to the standard. However, it is not easy to manage the signboards systematically due to the number of signboards that have been installed for a long time and frequently changing stores. In this study, we proposed a methodology for distinguishing signboards that deviated from the standard. To this end, the signboard was photographed using a stereo camera, and then the three-dimensional coordinates of the signboard were determined from the signboard image to calculate the signboard's horizontal and vertical dimensions to determine the signboard's specifications. In order to determine the interior and relative orientation parameters of the stereo camera, an outdoor three-dimensional building was used as the test field. Then, the image coordinates of four vertices of the signboard were extracted from the signboard image taken from about 15m ~ 22m distance using deep learning. After determining the signboard's three-dimensional coordinates by using the interior and relative orientation parameters of the stereo camera and the image coordinates of the four vertices of the signboard, the horizontal and vertical sizes of the signboard were calculated, resulting in an error of about 2.7cm on average. The specifications for the ten flat-type signboards showed that all of the horizontal sizes were compliant with the specifications, but the vertical sizes exceeded about 36.5cm on average. Through this, it was found that maintenance of flat-type signboards is needed overall.

• Journal of Biomedical Engineering Research
• /
• v.43 no.5
• /
• pp.353-360
• /
• 2022

Laryngeal disease harms quality of life, and laryngoscopy is critical in identifying causative lesions. This study extracts and analyzes using radiomics quantitative features from the lesion in laryngoscopy images and will fit and validate a classifier for finding meaningful features. Searching the region of interest for lesions not classified by the YOLOv5 model, features are extracted with radionics. Selected the extracted features are through a combination of three feature selectors, and three estimator models. Through the selected features, trained and verified two classification models, Random Forest and Gradient Boosting, and found meaningful features. The combination of SFS, LASSO, and RF shows the highest performance with an accuracy of 0.90 and AUROC 0.96. Model using features to select by SFM, or RIDGE was low lower performance than other things. Classification of larynx lesions through radiomics looks effective. But it should use various feature selection methods and minimize data loss as losing color data.