On pig farms, the highest mortality rate is observed among nursing piglets. To reduce this mortality rate, farmers need to carefully observe the piglets to prevent accidents such as being crushed and to maintain a proper body temperature. However, observing a large number of pigs individually can be challenging for farmers. Therefore, our aim was to detect the behavior of piglets and sows in real-time using deep learning models, such as YOLOv4-CSP and YOLOv7-E6E, that allow for real-time object detection. YOLOv4-CSP reduces computational cost by partitioning feature maps and utilizing Cross-stage Hierarchy to remove redundant gradient calculation. YOLOv7-E6E analyzes and controls gradient paths such that the weights of each layer learn diverse features. We detected standing, sitting, and lying behaviors in sows and lactating and starving behaviors in piglets, which indicate nursing behavior and movement to colder areas away from the group. We optimized the model parameters for the best object detection and improved reliability by acquiring data through experts. We conducted object detection for the five different behaviors. The YOLOv4-CSP model achieved an accuracy of 0.63 and mAP of 0.662, whereas the YOLOv7-E6E model showed an accuracy of 0.65 and mAP of 0.637. Therefore, based on mAP, which includes both class and localization performance, YOLOv4-CSP showed the superior performance. Such research is anticipated to be effectively utilized for the behavioral analysis of fattening pigs and in preventing piglet crushing in the future.

Recently, extreme terrorist attacks have frequently occurred around the world and are threatening the international community. It is no longer a safe zone for terrorism in our country. Therefore, domestic nuclear facilities as the highest level of national security facilities have established a physical protection system to protect facilities and lives against terrorist attacks. In addition, security search and access control are conducted for controlled items and unauthorized person. However, with the development of science and technology, disguised weapons or homemade explosives used in terrorism are becoming very sophisticated. Therefore, nuclear facilities need to strengthen security search of weapons or homemade explosives. Since these disguised weapons or homemade explosives are difficult to find only through security search, it is also necessary to actively identify unspecified people who possess disguised weapons or do abnormal behavior. For this reason, the “Abnormal Behavior Detection Method”, which is very effective in preemptive response to potential terrorist risks, has been introduced and operated in aviation security field. Korea Institute of Nuclear Nonproliferation and Control (KINAC) has established a “Practice Environment for Identifying Disguised Weapons” in 2020 for trainees to recognize the dangers of controlled items and to use for physical protection education. This Practice environment has not only the basic explanation of the controlled items of nuclear facilities, but also various actual disguised weapons were displayed. It also introduces actual terrorist incidents using homemade explosives such as attempted bombing of a cargo plane bound for Chicago and the Boston Marathon bombing. And then a model of the disguised explosives actually used is displayed and used for education. In addition, in 2022, the “Abnormal behavior detection method” education module was developed and used for physical protection education. In this module, the outline and introduction of the “Abnormal Behavior Detection Method” and “Behavior Detection Officer (BDOs)” are explained. In this way, the access control and security search system of nuclear facilities require the overall monitoring system, not only for dangerous goods but also for identification of persons possess and carrying them. This study describes the development of the Curriculum for “Disguised Weapon Identification” and “Abnormal Behavior Detection Method” to enhance the effectiveness of physical protection education.

This paper introduces unfamiliarity index (UFI) that calculated from the FFT results of the short term timeline acceleration responses. If this algorithm, which can detect an abnormal behavior from the maximum constant signal, is used to the terminal sensors of an structure, more accurate safety control criteria will be prepared efficiently.

All living organisms use memory and oblivion algorithms considering the estimated lifetime and the changes in the ambient environment. Because of the expected lifetime of a bridge is similar to the human’s one, if a bridge uses the same algorithm of human memory, the abnormal responses of the structure can be easily detected. This paper introduces unfamiliarity index (UFI) that calculated from the FFT results of the short term timeline acceleration responses. If this algorithm, which can detect an abnormal behavior from the maximum constant signal, is used to the terminal sensors of an structure, more accurate safety control criteria will be prepared efficiently.

The aims of this study were to investigated the occurrence of caffeine and carbamazepine in Nakdong river basin (8 mainstreams and 2 tributaries) and the behavior of caffeine and carbamazepine under drinking water treatment processes (conventional and advanced processes). The examination results showed that caffeine was detected at all sampling sites (5.4 ∼558.5 ng/L), but carbamazepine was detected at five sampling sites (5.1∼79.4 ng/L). The highest concentration level of caffeine and carbamazepine in the mainstream and tributaries in Nakdong river were Goryeong and Jinchun-cheon, respectively. These pharmaceutical products were completely removed when they were subject to conventional plus advanced processes of drinking water treatment processes. Conventional processes of coagulation, sedimentation and sand-filtration were not effective for their removal, while advanced processes of ozonation and biological activated carbon (BAC) filtration were effective. Among these pharmaceuticals, carbamazeoine was more subject to ozonation than caffeine.

온라인상에서 사용자의 개인정보를 불법적으로 취득, 악용하는 계정도용 문제는 금전적인 이득을 얻을 수 있는 MMORPG(Massively Multi-player Online Role Playing Games)에서 특히 빈번하게 발생하고 있다. 많은 사람들이 게임을 이용하여 심각한 피해로 이어질 수 있기 때문에 이에 대한 대책마련이 시급함에도 불구하고, 이를 예방하거나 탐지하는 기법에 대한 연구가 많이 부족한 실정이다. 본 연구에서는 온라인게임에서 발생했던 실제 계정도용 사례 분석을 통해 계정도용의 유형을 체계적으로 정의하고, 유형별로 계정도용을 분류하는 자동화된 탐지모델을 제안한다. 실 계정도용 사례를 분석한 결과 속전속결형, 신중형, 대담무쌍형의 3가지로 구분되었으며 이 분류 체계와 탐지모델을 국내 주요 온라인게임회사 중 한 곳에 적용하였다. 본 연구에서 제시한 유형별 탐지모델은 해킹의 유무만을 판정하던 기존의 모델보다 탐지에 있어서 향상된 성능을 보였다.

MMORPG (Massively Multiplayer Online Role Playing Game) 시장은 급격히 증가하고 있으며 더불어 많은 발전을 이루고 있다. 하지만 그와 동시에 많은 게임 피해사례들이 증가하고 그 사례 또한 매우 다양화되고 있다. 그 중에서도 '봇(Bots)'은 사용자의 조작 없이 자동으로 작동하면서 게임의 몰입도 뿐만 아니라 보안 측면에서도 맡은 영향을 주고 있다. 따라서 게임 제공 업체에서는 클라이언트 단에서 packet을 분석하여 봇를 분별하려 하지만 클라이언트 단에는 사용자의 조작이 용이하므로 그 정확성이 떨어진다. 본 논문에서는 게임 서버 내에서 얻을 수 있는 사용자의 행동 데이터를 분석함으로써 실제 사용자 및 봇의 행동 패턴을 모델링하고 이를 비교하여 봇 검출에 적용하는 방법을 제안한다. 이 방법을 이용하여 보다 향상된 비교 모델을 완성 하였다.