• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.505-507
• /
• 2021

Image recognition technology is a technology that recognizes an image object by using the generated feature descriptor and generates object feature points and feature descriptors that can compensate for the shape of the object to be recognized based on artificial intelligence technology, environmental changes around the object, and the deterioration of recognition ability by object rotation. The purpose of the present invention is to implement a power management framework required to increase profits and minimize damage to livestock farmers by preventing accidents that may occur due to the improvement of efficiency of the use of livestock house power and overloading of electricity by integrating and managing a power fire management device installed for analyzing a complex environment of power consumption and fire occurrence in a smart safety livestock house, and to develop and disseminate a safe and optimized intelligent smart safety livestock house.

• Annual Conference of KIPS
• /
• 2019.10a
• /
• pp.827-829
• /
• 2019

본 논문에서는 최근 소개된 구글 TPU 보드를 사용하여 AI 알고리듬을 적용하고 성능 분석을 통하여 TPU 를 통한 AI 에 기반한 영상처리 시스템의 구현 가능성을 검증 하고자 하였다. 구글 TPU 보드는 기계 학습에 특화된 Coral Dev 보드를 사용하였고. 수행하는 인공지능 알고리즘은 객체 인식 알고리즘인 SSD 알고리즘을 사용하였다. 이 후 동일한 알고리즘을 GPU 가 장착되어 있는 고성능 데스크탑과 처리속도를 비교하여, TPU 에 기반한 임베디드 AI 시스템의 활용 가능성을 검증 하였다.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.3
• /
• pp.45-52
• /
• 2022

A tangible sports game is an exercise game that uses sensors or cameras to track the user's body movements and to feel a sense of reality. Recently, VR indoor sports room systems installed to utilize tangible sports game for physical activity in schools. However, these systems primarily use screen-touch user interaction. In this research, we developed a V-Catch Vision system that uses AI image recognition technology to enable tracking of user movements in three-dimensional space rather than two-dimensional wall touch interaction. We also conducted a usability evaluation experiment to investigate the exercise effects of this system. We tried to evaluate quantitative exercise effects by measuring blood oxygen saturation level, the real-time ECG heart rate variability, and user body movement and angle change of Kinect skeleton. The experiment result showed that there was a statistically significant increase in heart rate and an increase in the amount of body movement when using the V-Catch Vision system. In the subjective evaluation, most subjects found the exercise using this system fun and satisfactory.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.3
• /
• pp.166-172
• /
• 2022

In this study, we photograph driving obstacle objects such as crosswalks, side spheres, manholes, braille blocks, partial ramps, temporary safety barriers, stairs, and inclined curb that hinder or cause inconvenience to the movement of the vulnerable using electric mobility aids. We develop an optimal AI model that classifies photographed objects and automatically recognizes them, and implement an algorithm that can efficiently determine obstacles in front of electric mobility aids. In order to enable object detection to be AI learning with high probability, the labeling form is labeled as a polygon form when building a dataset. It was developed using a Mask R-CNN model in Detectron2 framework that can detect objects labeled in the form of polygons. Image acquisition was conducted by dividing it into two groups: the general public and the transportation weak, and image information obtained in two areas of the test bed was secured. As for the parameter setting of the Mask R-CNN learning result, it was confirmed that the model learned with IMAGES_PER_BATCH: 2, BASE_LEARNING_RATE 0.001, MAX_ITERATION: 10,000 showed the highest performance at 68.532, so that the user can quickly and accurately recognize driving risks and obstacles.

• Annual Conference of KIPS
• /
• 2021.11a
• /
• pp.1188-1191
• /
• 2021

The function of the black box can be combined with a car to check the video through a cloud server, reduce the hassle of checking the video through a memory card, check the black box image in real time through a PC and smartphone, and check the user's Excel, brake operation status, and handle control record at the time of the accident. In addition, the goal was to accurately identify vehicle accidents and simplify accident handling through artificial intelligence object recognition of black box images using cloud services. Measures can be prepared to preserve images even if the black box itself loses, such as fire, flooding, or damage that occurs in an accident. It has been confirmed that the exact situation before and after the accident can be grasped immediately by providing object recognition and log recording functions under actual driving experimental conditions.

• Annual Conference of KIPS
• /
• 2022.05a
• /
• pp.637-640
• /
• 2022

3차원 물체의 탐지와 자세 추정은 실내외 환경에서 장면 이해, 로봇의 물체 조작 작업, 자율 주행, 증강 현실 등과 같은 다양한 응용 분야들에서 공통적으로 요구되는 매우 중요한 시각 인식 기술이다. 깊이 지도를 요구하는 기존 연구들과는 달리, 본 논문에서는 RGB 컬러 영상만을 이용해 미지의 물체들, 즉 3차원 CAD 모델을 가지고 있지 않은 새로운 물체들을 탐지해내고, 이들의 자세를 추정해낼 수 있는 새로운 신경망 모델을 제안한다. 제안 모델에서는 최근 빠른 속도로 발전하고 있는 깊이 추정 기술을 이용함으로써, 깊이 측정 센서 없이도 물체 자세 추정에 필요한 깊이 지도를 컬러 영상에서 구해낼 수 있다. 본 논문에서는 벤치마크 데이터 집합을 이용한 실험을 통해, 제안 모델의 유용성을 평가한다.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.25 no.3
• /
• pp.300-310
• /
• 2022

A vast amount of labeled data is required for deep neural network training. A typical strategy to improve the performance of a neural network given a training data set is to use data augmentation technique. The goal of this work is to offer a novel image augmentation method for improving object detection accuracy. An object in an image is removed, and a similar object from the training data set is placed in its area. An in-painting algorithm fills the space that is eliminated but not filled by a similar object. Our technique shows at most 2.32 percent improvements on mAP in our testing on a military vehicle dataset using the YOLOv4 object detector.

• Journal of the Korean Geosynthetics Society
• /
• v.23 no.2
• /
• pp.43-52
• /
• 2024

Recent abnormal climate conditions have increased the risk of slope collapses, which frequently result in significant loss of life and property due to the absence of early prediction and warning dissemination. In this paper, we develop a slope condition analysis system using IoT sensors and AI-based camera to assess the condition of slopes. To develop the system, we conducted hardware and firmware design for measurement sensors considering the ground conditions of slopes, designed AI-based image analysis algorithms, and developed prediction and warning solutions and systems. We aimed to minimize errors in sensor data through the integration of IoT sensor data and AI camera image analysis, ultimately enhancing the reliability of the data. Additionally, we evaluated the accuracy (reliability) by applying it to actual slopes. As a result, sensor measurement errors were maintained within 0.1°, and the data transmission rate exceeded 95%. Moreover, the AI-based image analysis system demonstrated nighttime partial recognition rates of over 99%, indicating excellent performance even in low-light conditions. Through this research, it is anticipated that the analysis of slope conditions and smart maintenance management in various fields of Social Overhead Capital (SOC) facilities can be applied.

• Smart Media Journal
• /
• v.10 no.1
• /
• pp.55-62
• /
• 2021

We applied machine learning of semi-supervised learning, transfer learning, and federated learning as examples of AI use cases that can be applied to the three major industries(Automobile industry, Energy industry, and AI/Healthcare industry) of Gwangju Metro-city, and established an ML strategy for AI services for the major industries. Based on the ML strategy of AI service, practical approaches are suggested, the semi-supervised learning approach is used for automobile image recognition technology, and the transfer learning approach is used for diabetic retinopathy detection in the healthcare field. Finally, the case of the federated learning approach is to be used to predict electricity demand. These approaches were tested based on hardware such as single board computer Raspberry Pi, Jaetson Nano, and Intel i-7, and the validity of practical approaches was verified.

• Journal of the Korean Society of Radiology
• /
• v.85 no.4
• /
• pp.769-779
• /
• 2024

Purpose To determine the pros and cons of an artificial intelligence (AI) model developed to diagnose acute rib fractures in chest CT images of patients with chest trauma. Materials and Methods A total of 1209 chest CT images (acute rib fracture [n = 1159], normal [n = 50]) were selected among patients with chest trauma. Among 1159 acute rib fracture CT images, 9 were randomly selected for AI model training. 150 acute rib fracture CT images and 50 normal ones were tested, and the remaining 1000 acute rib fracture CT images was internally verified. We investigated the diagnostic accuracy and errors of AI model for the presence and location of acute rib fractures. Results Sensitivity, specificity, positive and negative predictive values, and accuracy for diagnosing acute rib fractures in chest CT images were 93.3%, 94%, 97.9%, 82.5%, and 95.6% respectively. However, the accuracy of the location of acute rib fractures was low at 76% (760/1000). The cause of error in the diagnosis of acute rib fracture seemed to be a result of considering the scapula or clavicle that were in the same position (66%) or some ribs that were not recognized (34%). Conclusion The AI model for diagnosing acute rib fractures showed high accuracy in detecting the presence of acute rib fractures, but diagnosis of the exact location of rib fractures was limited.