• 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.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.5 no.1 s.9
• /
• pp.31-43
• /
• 2006

As of March 2005, the standard of traffic signal controllers became effective. The standard presents specifications and functions of a traffic signal controller which collects traffic information, sends it to the traffic control center, and controls traffic signal with adequate traffic signal timing provided by the traffic control center. Since the controllers by the previous standard lack parts compatibility and have different control functions and communication protocol, the maintenance cost has been increased. Also, some important functions like conflict detection have not worked out perfectly. To overcome these disadvantages, first of all, this standard secures hardware compatibility. Conflict detection method has been enhanced. Communication protocol to the traffic control center was included in the standard. With this standard, independent maintenance system and prompt treatment of hardware malfunctions becomes possible. Also, the unified intersection traffic control method will increase traffic safety.

• The Journal of the Korea Contents Association
• /
• v.18 no.3
• /
• pp.52-61
• /
• 2018

This research suggested the traffic signal timing calculation model for signal intersections based on sectional travel time. A detection system that collects sectional travel time data such as Urban Transport Information System(UTIS) is applied. This research developed the model to calculate saturation flow rate and demand volume from travel time information using a deterministic delay model. Moreover, this model could determine the traffic signal timings to minimize a delay based on Webster model using traffic demand volume. In micro simulation analysis using VISSIM and its API ComInterface, it checked the saturation conditions and determined the traffic signal timings to minimize the intersection delay. Recently, sectional vehicle detection systems are being installed in various projects, such as Urban Transportation Information System(UTIS) and Advanced Transportation Management System(ATMS) in Korea. This research has important contribution to apply the traffic information system to traffic signal operation sector.

• Journal of the Korea Society of Computer and Information
• /
• v.3 no.3
• /
• pp.17-22
• /
• 1998

As time goes on, that becomes an issue still more for truth from error of a seal in electronic settlement , or in important document in the field of image recognition. But on the other hand image treatment method of a seal have has the weakness until now. It makes indistinct distinction of part that light and darkness is changed sharply as the edge of things. So it has difficult that edge detection is extracted. In this paper, I investigated the pixel in a specific area by using enhanced smothing method and searched a value of frquent occurrence. The value of pixel is substituted and edge detection is extracted. After then it could be classified rightly according as viligence test is dynamically changed. I applied conventional of Yager's generated intersection operator among fuzzy logic operator in ART1 learning Algorithm. Application of suggested ART1 learning algorithm, it results in improved image recognition rate than a case of using the conventional ART1 algorithm

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.8 no.5
• /
• pp.26-37
• /
• 2009

This paper suggests a sensor's placement method for detecting the dilemma zone problem when real-time driver's safety service is provided at signalized intersections by multiple pointed traffic sensors using USN environments. For detecting the dangerous situations from vehicles accelerating through yellow intervals, red-light running and stopping abruptly like as dilemma zone problem, VISSIM(microscopic, behavior-based multi-purpose traffic simulation program) is used to perform a real-time multiple detection situation by changing the input data like as various inflow-volume, design speed change, driver perception and response time. As a result, the optimal interval of traffic sensors is 20~27m, and the initialized sensor location from stop-line is different according to road design speed. Moreover, the pattern of detection about dilemma zone is also different according to inflow-volumes. This paper shows that the method is useful to evaluate the sensor's placement problem based on micro-simulation and the results can be used as the basic research for USN services.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.14 no.1
• /
• pp.1-12
• /
• 2015

Intrusive Vehicle Detectors have excellent detection performance compared to other types of detector, but disadvantages of high installation and maintenance costs, short life time due to greater damage to roads and paving materials. In contrast, Non-Intrusive Vehicle Detectors attached to the stationary pole have advantages because it does not damage the road surface and easy and less expensive to maintain. Despite these advantages, Non-Intrusive type detectors are still not been widely used in traffic signal control systems because of the low detection performance. In this study, a FMCW(Frequency Modulated Continuous Wave) radar Vehicle Detector was designed as an alternative detector for the signalized intersection, and the performance evaluation was presented by purpose applicability.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.137-137
• /
• 2023

해양쓰레기는 해안의 심미적 가치 저하뿐만 아니라 생태계 파괴, 유령 어업에 따른 수산업 피해 등의 사회적·환경적 문제를 발생시키며, 그중 70% 이상은 육상 기인으로 플라스틱 및 기타 쓰레기가 주를 이루는 해외와 달리 국내의 경우 다량의 초목류를 포함하고 있다. 다양한 부유쓰레기에 대한 기존의 해양쓰레기량 추정의 한계와 하천·하구 쓰레기 수거의 효율화를 위해 해양으로 유입되는 부유쓰레기 방지를 위한 실효성 있는 대책 수립이 필요한 실정이다. 본 연구는 해양 유입 전 하천의 차단시설에 차집된 부유쓰레기의 수거 효율화 및 지속가능한 해양쓰레기 데이터 구축을 위해 AI기반의 기술을 통해 부유쓰레기 성상 분석 기법(Object Detection)과 차집량 분석 기법(Semantic Segmentation)을 활용하였다. 실제와 유사한 데이터 수집을 위해 다양한 하천 환경(정수조, 소하천, 급경사수로)에 대해 탁도(녹조, 유사), 광량, 쓰레기형상, 초목류 함량, 날씨(소하천), 유속(급경사수로) 등의 실험조건에 대하여 해양쓰레기 분류 기준 및 통계를 바탕으로 부유쓰레기 종류 선정하여 학습을 위한 데이터를 수집하였다. 학습 목적에 따라 구분하여 라벨링(Bounding box, Polygon)을 수행하고, 각 분석 기법별 전이학습을 통해 Phase 1(정수조), Phase 2(소하천), Phase 3(급경사수로) 순서로 모델을 고도화하였다. 성상 분석을 위해 YOLO v4를 활용하여 Train, Test DataSet(9:1)을 구성하고 학습 및 평가는 Iteration마다의 mAP, loss 값을 통해 비교하였으며, 학습 Phase에 따라 모델 고도화로 Test Set의 mAP 값이 성상별로 높아짐을 확인하였으며, 차집량 분석을 위해 Unet을 활용하여 Train, Test, Validation DataSet(8.5:1:0.5)을 구성하고 epoch별 IoU(intersection over Union), F1-score, loss 값을 비교하여 정성적, 정량적 평가 모두 Phase 3에서 가장 높은 성능을 확인하였다. 향후 하천 환경에서의 다양한 영양인자별 분석을 통해 주요 영향인자 도출 및 Hyper Parameter 최적화를 통한 모델 고도화로 인해 활용성이 높아질 것으로 판단된다.

• Advances in concrete construction
• /
• v.16 no.1
• /
• pp.69-78
• /
• 2023

This investigation delves into the adverse repercussions stemming from the impact of arsenic on steel pipes concealed within soil designated for rice cultivation. Simultaneously, the study aims to ascertain effective techniques for detecting arsenic in the soil and to provide strategies for mitigating the corrosion of steel pipes. The realm of nanotechnology presents promising avenues for addressing the intricate intersection of renewable energy, oil, and environmental pollution from a novel perspective. Nanostructured materials, characterized by distinct chemical and physical attributes, unveil novel pathways for pioneering materials that exert a substantial impact across diverse realms of food production, storage, packaging, and quality control. Within the scope of the food industry, the scope of nanotechnology encompasses processes, storage methodologies, packaging paradigms, and safeguards to ensure the safety of consumables. Of particular note, silver nanoparticles, in addition to their commendable antibacterial efficacy, boast anti-fungal and anti-inflammatory prowess, environmental compatibility, minimal irritability and allergenicity, resilience to microbial antagonism, thermal stability, and robustness. Confronting the pressing issue of arsenic contamination within both environmental settings and the food supply is of paramount importance to preserve public health and ecological equilibrium. In response, this study introduces detection kits predicated upon silver nanoparticles, providing an expeditious and economically feasible avenue for identifying arsenic concentrations ranging from 0.5 to 3 ppm within rice. Subsequent quantification employs Hydride Atomic Absorption Spectroscopy (HG-AAS), which features a detection threshold of 0.05 ㎍/l. A salient advantage inherent in the HG-AAS methodology lies in its capacity to segregate analytes from the sample matrix, thereby significantly reducing instances of spectral interference. Importantly, the presence of arsenic in the soil beneath rice cultivation establishes a causative link to steel pipe corrosion, with potential consequences extending to food contamination-an intricate facet embedded within the broader tapestry of renewable energy, oil, and environmental pollution.

• Clinical Endoscopy
• /
• v.55 no.3
• /
• pp.390-400
• /
• 2022

Background/Aims: Previous artificial intelligence (AI) models attempting to segment gastric intestinal metaplasia (GIM) areas have failed to be deployed in real-time endoscopy due to their slow inference speeds. Here, we propose a new GIM segmentation AI model with inference speeds faster than 25 frames per second that maintains a high level of accuracy. Methods: Investigators from Chulalongkorn University obtained 802 histological-proven GIM images for AI model training. Four strategies were proposed to improve the model accuracy. First, transfer learning was employed to the public colon datasets. Second, an image preprocessing technique contrast-limited adaptive histogram equalization was employed to produce clearer GIM areas. Third, data augmentation was applied for a more robust model. Lastly, the bilateral segmentation network model was applied to segment GIM areas in real time. The results were analyzed using different validity values. Results: From the internal test, our AI model achieved an inference speed of 31.53 frames per second. GIM detection showed sensitivity, specificity, positive predictive, negative predictive, accuracy, and mean intersection over union in GIM segmentation values of 93%, 80%, 82%, 92%, 87%, and 57%, respectively. Conclusions: The bilateral segmentation network combined with transfer learning, contrast-limited adaptive histogram equalization, and data augmentation can provide high sensitivity and good accuracy for GIM detection and segmentation.

• The Journal of Korea Robotics Society
• /
• v.11 no.1
• /
• pp.1-8
• /
• 2016

In this paper, we propose a new algorithm of the guidance line extraction for autonomous agricultural robot based on vision camera in paddy field. It is the important process for guidance line extraction which finds the central point or area of rice row. We are trying to use the central region data of crop that the direction of rice leaves have convergence to central area of rice row in order to improve accuracy of the guidance line. The guidance line is extracted from the intersection points of extended virtual lines using the modified robust regression. The extended virtual lines are represented as the extended line from each segmented straight line created on the edges of the rice plants in the image using the Hough transform. We also have verified an accuracy of the proposed algorithm by experiments in the real wet paddy.