Search

Recently, with the rapid increase in the popularity of car-sharing services, there has been a rise in cases where the license holder and the actual driver differ, leading to an increase in traffic accidents caused by drivers lacking sufficient experience. As an innovative solution, this paper proposes an identity verification system that real-time compares registered license holders. The system utilizes an advanced application based on ArcFace technology, incorporating tools like the RetinaFace library and finely tuned ArcFace models. An Android app is integral to effectively and reliably authenticate the driver's identity. This approach addresses safety concerns in car-sharing services by ensuring only qualified, registered drivers are at the wheel, eliminating unqualified or inexperienced drivers. It is anticipated to significantly reduce accidents caused by such drivers. This technological advancement in driver authentication is poised to make car-sharing services safer and more reliable, benefiting users and the local community. [ABSTRACT FROM AUTHOR]

There were more than 1.6 million CCTVs operated by local governments and public organizations as of early 2023. CCTV is very effective in helping people live safely and responding to emergencies. However, current public CCTVs often have inadequate video input quality due to limited viewing distance, errors in object recognition depth, and obstructions on the CCTV screen such as leaves, wires, and street lights, which directly lead to gradient vanishing. In addition, CCTV video information must be understood as spatial information for real-time object detection, identification, and tracking. Therefore, this paper studies the quality of input images by processing interactive replacement events to improve the reliability of image analysis. In addition, after analyzing the input image, virtual lines are applied to the spatial information system to measure the movement speed of the object and track the optimal path based on artificial intelligence. The results of this study will improve the safe living environment for the public [ABSTRACT FROM AUTHOR]

Local governments in Korea are installing and operating emergency bells. The emergency bell must transmit the user's emergency rescue call or crisis situation to the control center or police station. However, in the case of local government public emergency bells, in addition to situation-specific delivery, two-way emergency response is required to respond to emergency situations, and emergency support must be available even when personal information protection is necessary depending on the situation. Additionally, it must be possible to provide accurate answers to citizens' questions or requests. In this paper, we connect Whisper and ChatGPT (Generative Pre-trained Transformer) systems to local government emergency bells, respond to emergency situation communication analysis through artificial intelligence video analysis, voice and text recognition and learning, and provide emergency response to more accurate responses. I want to study warning systems. This study is expected to provide a solution to improve the quality of life of the people [ABSTRACT FROM AUTHOR]

National R&D information includes various contents generated by national R&D projects and is serviced by the National Science and Technology Knowledge Information Service (NTIS). NTIS provides a search engine-based search method. This method provides search results by sorting them based on similarity ranking, and there is a structural limitation in that users have to search the initial search results several times or filter them to find the information they want. To address these limitations, this paper describes the core model of the R&D recommendation information service that recommends content by analyzing user usage patterns and characteristics of content. This model is a hybrid model that combines collaborative filtering, content-based filtering, and deep learning models, which are widely used in the field of recommending information, and is intended to further utilize preferred/non-preferred pattern analysis techniques derived by analyzing the usage rate of recommended information for each user. Through this, users are expected to receive improved recommendation results [ABSTRACT FROM AUTHOR]

The hot forging compound injection technique is one of the processes in hot forging where various metal materials are heated and pressure is applied to deform the shape of the object, and this technology is utilized in various industries, including the automotive, aerospace, and machinery sectors. The conventional hot forging compound injection technique has some issues as it mainly relies on manual labor, leading to several problems such as metal material waste and increased operating costs. In this paper, we aim to overcome these issues and limitations of the conventional hot forging compound injection method by developing an artificial intelligence-based technique for measuring the injection amount and coating quality. In detail, Generative Adversarial Network(GAN) is employed for image augmentation, You Only Look Once(YOLO) for injection pattern detection, Linear Regression for injection volume measurement, and Convolutional Neural Network(CNN) along with Support Vector Machine (SVM)for coating quality analysis. [ABSTRACT FROM AUTHOR]

Pancreatic Ductal Adenocarcinoma (PDAC) is the most common and deadly form of pancreatic cancer. Currently, histopathological diagnosis and prognosis of PDAC are time-consuming and labor-intensive for pathologists. Recent advances in pathological AI research aim to alleviate this. We accumulated training data, distinguishing PDAC areas in Whole Slide Images (WSIs) based on medical findings. Using this data, we trained a deep convolutional neural network for supervised learning to automatically interpret PDAC areas. The AI model achieved high Dice scores and, by visualizing the segmentation results of the predicted histological images, validated that PDAC diagnosis and identification of associated regions are automatically possible, similar to pathologists. Additionally, the AI model, which showed high specificity, suggests its potential as a co-pilot for pathological diagnosis and annotation. [ABSTRACT FROM AUTHOR]