• Journal of Digital Convergence
• /
• v.20 no.5
• /
• pp.39-46
• /
• 2022

As the link between traffic accidents and social and economic losses has been confirmed, there is a growing interest in developing safety policies based on crash data and a need for countermeasures to reduce severe crash outcomes such as severe injuries and fatalities. In this study, we select Daejeon city where the relative proportion of fatal crashes is high, as a case study region and focus on the severity of pedestrian crashes. After a series of data manipulation process, we run machine learning algorithms for the optimal model selection and variable identification. Of nine algorithms applied, AdaBoost and Random Forest (ensemble based ones) outperform others in terms of performance metrics. Based on the results, we identify major influential factors (i.e., the age of pedestrian as 70s or 20s, pedestrian crossing) on pedestrian crashes in Daejeon, and suggest them as measures for reducing severe outcomes.

• IEIE Transactions on Smart Processing and Computing
• /
• v.5 no.4
• /
• pp.256-266
• /
• 2016

Concurrently detected and annotated abnormal events can have a significant impact on surveillance systems. By considering the specific domain of pedestrian trajectories, this paper presents two main contributions. First, as introduced in much of the work on trajectory-based anomaly detection in the literature, only information about pedestrian paths, such as direction and speed, is considered. Differing from previous work, this paper proposes a framework that deals with additional types of trajectory-based anomalies. These abnormal events take places when a person enters prohibited areas. Those restricted regions are constructed by an online learning algorithm that uses surrounding information, including detected pedestrians and background scenes. Second, a simple data-boosting technique is introduced to overcome a lack of training data; such a problem particularly challenges all previous work, owing to the significantly low frequency of abnormal events. This technique only requires normal trajectories and fundamental information about scenes to increase the amount of training data for both normal and abnormal trajectories. With the increased amount of training data, the conventional abnormal trajectory classifier is able to achieve better prediction accuracy without falling into the over-fitting problem caused by complex learning models. Finally, the proposed framework (which annotates tracks that enter prohibited areas) and a conventional abnormal trajectory detector (using the data-boosting technique) are integrated to form a united detector. Such a detector deals with different types of anomalous trajectories in a hierarchical order. The experimental results show that all proposed detectors can effectively detect anomalous trajectories in the test phase.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.18 no.4
• /
• pp.58-70
• /
• 2019

In the case of a pedestrian traffic accident, it has a large-scale danger directly connected by a fatal accident at the time of the accident. The domestic ITS is not used for intelligent risk classification because it is used only for collecting traffic information despite of the construction of good quality traffic infrastructure. The CNN based pedestrian detection classification model, which is a major component of the proposed system, is implemented on an embedded system assuming that it is installed and operated in a restricted environment. A new model was created by improving YOLO's artificial neural network, and the real-time detection speed result of average accuracy 86.29% and 21.1 fps was shown with 20,000 iterative learning. And we constructed a protocol interworking scenario and implementation of a system that can connect with the ITS. If a pedestrian accident prevention system connected with ITS will be implemented through this study, it will help to reduce the cost of constructing a new infrastructure and reduce the incidence of traffic accidents for pedestrians, and we can also reduce the cost for system monitoring.

• Wind and Structures
• /
• v.11 no.1
• /
• pp.51-70
• /
• 2008

CFD is applied to evaluate pedestrian wind comfort at outdoor platforms in a high-rise apartment building. Model validation is focused on generic building sub-configurations that are obtained by decomposition of the actual complex building geometry. The comfort study is performed during the design stage, which allows structural design changes to be made for wind comfort improvement. Preliminary simulations are performed to determine the effect of different design modifications. A full wind comfort assessment study is conducted for the final design. Structural remedial measures for this building, aimed at reducing pressure short-circuiting, appear to be successful in bringing the discomfort probability estimates down to acceptable levels. Finally, the importance of one of the main sources of uncertainty in this type of wind comfort studies is illustrated. It is shown that the uncertainty about the terrain roughness classification can strongly influence the outcome of wind comfort studies and can lead to wrong decisions. This problem is present to the same extent in both wind tunnel and CFD wind comfort studies when applying the same particular procedure for terrain relation contributions as used in this paper.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.5
• /
• pp.1010-1018
• /
• 2009

An inertial navigation system for pedestrian position tracking is proposed, where the position is computed using inertial sensors mounted on shoes. Inertial navigation system(INS) errors increase with time due to inertial sensor errors, and therefore it needs to reset errors frequently. During normal walking, there is an almost periodic zero velocity instance when a foot touches the floor. Using this fact, estimation errors are reduced and this method is called the zero velocity updating algorithm. When implementing this zero velocity updating algorithm, it is important to know when is the zero velocity interval. The gait states are modeled as a Markov process and each state is estimated using the hidden Markov model smoother. With this gait estimation, the zero or nearly zero velocity interval is more accurately estimated, which helps to reduce the position estimation error.

• ETRI Journal
• /
• v.45 no.5
• /
• pp.862-873
• /
• 2023

Determining whether an autonomous self-driving agent is in the middle of an intersection can be extremely difficult when relying on visual input taken from a single camera. In such a problem setting, a wider range of views is essential, which drives us to use three cameras positioned in the front, left, and right of an agent for better intersection recognition. However, collecting adequate training data with three cameras poses several practical difficulties; hence, we propose using data collected from one camera to train a three-camera model, which would enable us to more easily compile a variety of training data to endow our model with improved generalizability. In this work, we provide three separate fusion methods (feature, early, and late) of combining the information from three cameras. Extensive pedestrian-view intersection classification experiments show that our feature fusion model provides an area under the curve and F1-score of 82.00 and 46.48, respectively, which considerably outperforms contemporary three- and one-camera models.

• Journal of the Semiconductor & Display Technology
• /
• v.23 no.1
• /
• pp.61-64
• /
• 2024

As research on autonomous driving technology becomes more active, various studies on signal recognition of traffic lights are also being conducted. When recognizing traffic lights with different purposes and shapes, such as pedestrian traffic lights, vehicle-only traffic lights, and right-turn traffic lights, existing classification methods may cause misrecognition problems. Therefore, in this study, we studied a model that allows accurate signal recognition by subdividing the classification of signals according to the purpose and type of traffic lights. A signal recognition model was created by classifying traffic lights according to their shape and purpose into horizontal, vertical, right turn, etc., and by comparing them with the existing signal recognition model based on YOLOv5, it was confirmed that more correct and accurate recognition was possible.

• Journal of the Architectural Institute of Korea Planning & Design
• /
• v.34 no.5
• /
• pp.71-78
• /
• 2018

This study seeks to identify elements of spatial planning for areas surrounding children's parks in order to improve child safety. It will identify different aspects of child safety and extract elements of safety planning for the areas surrounding children's parks from a literature review of both domestic and international research. The study classifies the space between one's residence and the children's park as either a "means of access", such as pedestrian walkways and streets, or a "boundary", which consists of the entrance, exit, and fences; then, the derived safety planning elements were categorized in accordance with this classification. In order to ensure the validity of the planning elements, an expert survey was conducted of environment planners who specialize in the palnning and designing of residential area as well as government employees who are in direct charge of managing children's parks. The survey findings were as follow. First, the pedestrian walkway(means of access) near the park is the most crucial factor in regards to the safety of children, followed by the entrance and exit(boundary), fency(boundary), and streets (means of access), in descending order of importance. Thus, improving the safety of the pedestrian walkway should be considered first and foremost, and it should precede improving the surrounding streets. Second, an investigation of the need for safety devices near the children's park showed that securing visibility, through the installation of an illegal parking prevention device, is imperative. Illegal parking near children's parks poses a grave risk to pedestrian safety and demands immediate action. Furthermore, a section of streets within 300m of the park entrance should be designated as a children protection zone, in addition to the designation of school zones near elementary schools.

• Journal of Internet Computing and Services
• /
• v.20 no.4
• /
• pp.91-102
• /
• 2019

In autonomous driving systems, the ability to classify pedestrians in images captured by cameras is very important for pedestrian safety. In the past, after extracting features of pedestrians with HOG(Histogram of Oriented Gradients) or SIFT(Scale-Invariant Feature Transform), people classified them using SVM(Support Vector Machine). However, extracting pedestrian characteristics in such a handcrafted manner has many limitations. Therefore, this paper proposes a method to classify pedestrians reliably and effectively using CNN's(Convolutional Neural Network) deep features and transfer learning. We have experimented with both the fixed feature extractor and the fine-tuning methods, which are two representative transfer learning techniques. Particularly, in the fine-tuning method, we have added a new scheme, called M-Fine(Modified Fine-tuning), which divideslayers into transferred parts and non-transferred parts in three different sizes, and adjusts weights only for layers belonging to non-transferred parts. Experiments on INRIA Person data set with five CNN models(VGGNet, DenseNet, Inception V3, Xception, and MobileNet) showed that CNN's deep features perform better than handcrafted features such as HOG and SIFT, and that the accuracy of Xception (threshold = 0.5) isthe highest at 99.61%. MobileNet, which achieved similar performance to Xception and learned 80% fewer parameters, was the best in terms of efficiency. Among the three transfer learning schemes tested above, the performance of the fine-tuning method was the best. The performance of the M-Fine method was comparable to or slightly lower than that of the fine-tuningmethod, but higher than that of the fixed feature extractor method.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.13 no.5
• /
• pp.35-49
• /
• 2014

This paper proposes the on-road object detection and classification algorithm by using a detection system consisting of only laser scanners. Each sensor data acquired by the laser scanner is fused with a grid map and the measurement error and spot spaces are corrected using a labeling method and dilation operation. Fuzzy method which uses the object information (length, width) as input parameters can classify the objects such as a pedestrian, bicycle and vehicle. In this way, the accuracy of the detection system is increased. Through experiments for some scenarios in the real road environment, the performance of the proposed detection and classification system for the actual objects is demonstrated through the comparison with the actual information acquired by GPS-RTK.