• Journal of Intelligence and Information Systems
• /
• v.28 no.1
• /
• pp.89-106
• /
• 2022

Recently, the amount of video data collected from smartphones, CCTVs, black boxes, and high-definition cameras has increased rapidly. According to the increasing video data, the requirements for analysis and utilization are increasing. Due to the lack of skilled manpower to analyze videos in many industries, machine learning and artificial intelligence are actively used to assist manpower. In this situation, the demand for various computer vision technologies such as object detection and tracking, action detection, emotion detection, and Re-ID also increased rapidly. However, the object detection and tracking technology has many difficulties that degrade performance, such as re-appearance after the object's departure from the video recording location, and occlusion. Accordingly, action and emotion detection models based on object detection and tracking models also have difficulties in extracting data for each object. In addition, deep learning architectures consist of various models suffer from performance degradation due to bottlenects and lack of optimization. In this study, we propose an video analysis system consists of YOLOv5 based DeepSORT object tracking model, SlowFast based action recognition model, Torchreid based Re-ID model, and AWS Rekognition which is emotion recognition service. Proposed model uses single-linkage hierarchical clustering based Re-ID and some processing method which maximize hardware throughput. It has higher accuracy than the performance of the re-identification model using simple metrics, near real-time processing performance, and prevents tracking failure due to object departure and re-emergence, occlusion, etc. By continuously linking the action and facial emotion detection results of each object to the same object, it is possible to efficiently analyze videos. The re-identification model extracts a feature vector from the bounding box of object image detected by the object tracking model for each frame, and applies the single-linkage hierarchical clustering from the past frame using the extracted feature vectors to identify the same object that failed to track. Through the above process, it is possible to re-track the same object that has failed to tracking in the case of re-appearance or occlusion after leaving the video location. As a result, action and facial emotion detection results of the newly recognized object due to the tracking fails can be linked to those of the object that appeared in the past. On the other hand, as a way to improve processing performance, we introduce Bounding Box Queue by Object and Feature Queue method that can reduce RAM memory requirements while maximizing GPU memory throughput. Also we introduce the IoF(Intersection over Face) algorithm that allows facial emotion recognized through AWS Rekognition to be linked with object tracking information. The academic significance of this study is that the two-stage re-identification model can have real-time performance even in a high-cost environment that performs action and facial emotion detection according to processing techniques without reducing the accuracy by using simple metrics to achieve real-time performance. The practical implication of this study is that in various industrial fields that require action and facial emotion detection but have many difficulties due to the fails in object tracking can analyze videos effectively through proposed model. Proposed model which has high accuracy of retrace and processing performance can be used in various fields such as intelligent monitoring, observation services and behavioral or psychological analysis services where the integration of tracking information and extracted metadata creates greate industrial and business value. In the future, in order to measure the object tracking performance more precisely, there is a need to conduct an experiment using the MOT Challenge dataset, which is data used by many international conferences. We will investigate the problem that the IoF algorithm cannot solve to develop an additional complementary algorithm. In addition, we plan to conduct additional research to apply this model to various fields' dataset related to intelligent video analysis.

• Journal of the Korean Society of Radiology
• /
• v.17 no.6
• /
• pp.851-857
• /
• 2023

The purpose of this study is to investigate the effect of changes in collimation size and sub ROI on exposure index(EI) in hand radiography, present collimation size and EI suitable for average hand size of Koreans, and present the effect of changes in sub ROI on EI. The subjects of this study were hand-wrist phantom, and the exposure conditions were set to 55 kVp, 125, mA, and 6.25 mAs, and source to image receptor distance was applied to 110 cm. Based on the vendor recommended sub-ROI (18.7" × 18.7", 8" × 10", 8" × 7.4", 6" × 7.4")and the textbook's recommended sub-ROI 8" × 10", each obtaining 30 images, and comparing the EI shown in the equipment. The EI according to the change in the size of the collimation were 1663.7±4.52, 8"×10" is 1489.1±4.49, 8"×7.4" is 1716.9±3.00, 6"×7.4" is 168.7±3.66 for each EI, and the average value of each value was statistically significant. The average EI according to the sub ROI change was 1489.1±4.49 for SS, LS was 1694.8±5.19 for AEC, 2052.9±5.96, VR was 1548.3±3.20, and HR was 1663.2±4.33. The appropriate field size considering the hand size of Koreans was found to be 8"×7.4". In addition, when the field size increases based on the generally known field size (8"×10") during hand radiography, the EI value changes from a maximum of 15% to a minimum of 11%, and the sub ROI shape based on sub ROI 'SS' Depending on the change, the EI value increased from a maximum of 37% to a minimum of 3%.

• Sim-seong Yeon-gu
• /
• v.28 no.1
• /
• pp.50-87
• /
• 2013

C. G. Jung believed that universal and basic condition of human's Unconscious comes out from Märchen or mythology. We can easily experience these universality of human nature in dreams. Therefore, It is very important to interpret mythogens that appear in myths and märchen in analytical psychology to understand these 'big dreams' which could be seen in clinical practice. As I was interested in interpreting myths in analytic psychology, I tried to find universality of archetypes in Korea's traditional folk tales and took note of the birth myth of Hyeokgeose, the founder of Shilla dynasty, while examining the chater of the Unsual in history in the Heritage of the Three Kingdoms. Shilla was founded earlier than two other countries, but it was located in the very south of the Korean Peninsula, and it was behind times in politically, militarily, and culturally compare to Goguryeo and Baekje. However, Shilla achieved unifying the Three Kingdoms and it lasted 1000 years, the longest unified history in Korean history. I tried to examine archetypes in the birth myth if there are any backgrounds that are related to finding a Shilla Kingdom. It is noted that myth of the founder of Korean Peninsula's small Kingdom Shilla has complete story from before the birth to birth, birth of spouse, growth, marriage, accession, governing, death, after death, and succession. Symbols such as numbers 1, 3, 5, 6, 7, 13 and 61, various azimuthes including north, west, south, east, and central, animals like tiger, white horse, hen, dragon, phoenix, and snakes, natures like main symbol egg, rock, gourd, lightening, spring water, stream, tree, forest, mountain, iron and goddess-image like seon-do Holy Mother gradually appears in the myth. These symbols could show a meaning of human experience such as birth of Conscious, growth and development of paternal and maternal love, and story of regeneration and extinction. Moreover, It could be seen as these progress eternally continues in next generation. I have found out that a word, a sentence or stories that looks meaningless in myth revealed its true symbolical meaning. In addition, interaction between Unconscious and Conscious repeats in different forms, and expressed in layered.

• The Journal of Korean Society for Radiation Therapy
• /
• v.30 no.1_2
• /
• pp.201-208
• /
• 2018

Purpose : The purpose of this study is to compare and analyze the set-up errors using the Combine IGRT with ExacTrac and CBCT phased in the treatment of Stereotatic Body Radiotherapy. Methods and materials : Patient who were treated Stereotatic Body Radiotherapy in the ulsan university hospital from May 2014 to november 2017 were classified as treatment area three brain, nine spine, three pelvis. First using ExacTrac Set-up error calibrated direction of Lateral(Lat), Longitudinal(Lng), Vertical(Vrt), Roll, Pitch, Yaw, after applied ExacTrac moving data in addition to use CBCT and set-up error calibrated direction of Lat, Lng, Vrt, Rotation(Rtn). Results : When using ExacTrac, the error in the brain region is Lat $0.18{\pm}0.25cm$, Lng $0.23{\pm}0.04cm$, Vrt $0.30{\pm}0.36cm$, Roll $0.36{\pm}0.21^{\circ}$, Pitch $1.72{\pm}0.62^{\circ}$, Yaw $1.80{\pm}1.21^{\circ}$, spine Lat $0.21{\pm}0.24cm$, Lng $0.27{\pm}0.36cm$, Vrt $0.26{\pm}0.42cm$, Roll $1.01{\pm}1.17^{\circ}$, Pitch $0.66{\pm}0.45^{\circ}$, Yaw $0.71{\pm}0.58^{\circ}$, pelvis Lat $0.20{\pm}0.16cm$, Lng $0.24{\pm}0.29cm$, Vrt $0.28{\pm}0.29cm$, Roll $0.83{\pm}0.21^{\circ}$, Pitch $0.57{\pm}0.45^{\circ}$, Yaw $0.52{\pm}0.27^{\circ}$ When CBCT is performed after the couch movement, the error in brain region is Lat $0.06{\pm}0.05cm$, Lng $0.07{\pm}0.06cm$, Vrt $0.00{\pm}0.00cm$, Rtn $0.0{\pm}0.0^{\circ}$, spine Lat $0.06{\pm}0.04cm$, Lng $0.16{\pm}0.30cm$, Vrt $0.08{\pm}0.08cm$, Rtn $0.00{\pm}0.00^{\circ}$, pelvis Lat $0.06{\pm}0.07cm$, Lng $0.04{\pm}0.05cm$, Vrt $0.06{\pm}0.04cm$, Rtn $0.0{\pm}0.0^{\circ}$. Conclusion : Combine IGRT with ExacTrac in addition to CBCT during Stereotatic Body Radiotherapy showed that it was possible to reduce the set-up error of patients compared to single ExacTrac. However, the application of Combine IGRT increases patient set-up verification time and absorption dose in the body for image acquisition. Therefore, depending on the patient's situation that using Combine IGRT to reduce the patient's set-up error can increase the radiation treatment effectiveness.