• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.22 no.4
• /
• pp.97-113
• /
• 2023

Swift and appropriate responses in emergency situations like objects falling on the road can bring convenience to road users and effectively reduces secondary traffic accidents. In Korea, current intelligent transportation system (ITS)-based detection systems for emergency road situations mainly rely on loop detectors and CCTV cameras, which only capture road data within detection range of the equipment. Therefore, a new detection method is needed to identify emergency situations in spatially shaded areas that existing ITS detection systems cannot reach. In this study, we propose a ResNet-based algorithm that detects and classifies emergency situations from vehicle camera footage. We collected front-view driving videos recorded on Korean highways, labeling each video by defining the type of emergency, and training the proposed algorithm with the data.

• Journal of Broadcast Engineering
• /
• v.14 no.4
• /
• pp.409-418
• /
• 2009

This paper proposes a method to estimate the flow speed of pedestrians in surveillance videos. In the proposed method, the average moving speed of pedestrians is measured by estimating the size of real-world motion from the observed motion vectors. For this purpose, a pixel-to-meter conversion factor is introduced which is calculated from camera parameters. Also, the height information, which is missing because of camera projection, is predicted statistically from simulation experiments. Compared to the previous works for flow speed estimation, our method can be applied to various camera views because it separates scene parameters explicitly. Experiments are performed on both simulation image sequences and real video. In the experiments on simulation videos, the proposed method estimated the flow speed with average error of about 0.08m/s. The proposed method also showed promising results for the real video.

• Smart Media Journal
• /
• v.13 no.2
• /
• pp.68-84
• /
• 2024

This paper describes the development of a system and algorithms for high-quality welfare services by recognizing behavior development indicators (activity, sociability, danger) in children aged 0 to 2 years old using action recognition technology. Action recognition targeted 11 behaviors from lying down in 0-year-olds to jumping in 2-year-olds, using data directly obtained from actual videos provided for research purposes by three nurseries in the Gwangju and Jeonnam regions. A dataset of 1,867 actions from 425 clip videos was built for these 11 behaviors, achieving an average recognition accuracy of 97.4%. Additionally, for real-world application, the Edge Video Analyzer (EVA), a behavior analysis device, was developed and implemented with a region-specific random frame selection-based PoseC3D algorithm, capable of recognizing actions in real-time for up to 30 people in four-channel videos. The developed system was installed in three nurseries, tested by ten childcare teachers over a month, and evaluated through surveys, resulting in a perceived accuracy of 91 points and a service satisfaction score of 94 points.

• Journal of Internet Computing and Services
• /
• v.9 no.3
• /
• pp.169-178
• /
• 2008

In this paper, we propose a video summarization system which is based on activity in video acquired by multiple non-overlapping cameras for wide-area surveillance. The proposed system separates persons by time-independent background removal and detects activities of the segmented persons by their motions. In this paper, we extract eleven activities based on whose direction the persons move to and consider a key-frame as a frame which contains a meaningful activity. The proposed system summarizes based on activity-based key-frames and controls an amount of summarization according to an amount of activities. Thus the system can summarize videos by camera, time, and activity.

• IEIE Transactions on Smart Processing and Computing
• /
• v.6 no.4
• /
• pp.237-245
• /
• 2017

This paper presents a fire detection algorithm using color and motion models from video sequences. The proposed method detects change in color and motion of overall regions for detecting fire, and thus, it can be implemented in both fixed and pan/tilt/zoom (PTZ) cameras. The proposed algorithm consists of three parts. The first part exploits color models of flames and smoke. The candidate regions in the video frames are extracted with the hue-saturation-value (HSV) color model. The second part models the motion information of flames and smoke. Optical flow in the fire candidate region is estimated, and the spatial-temporal distribution of optical flow vectors is analyzed. The final part accumulates the probability of fire in successive video frames, which reduces false-positive errors when fire-like color objects appear. Experimental results from 100 fire videos are shown, where various types of smoke and flames appear in indoor and outdoor environments. According to the experiments and the comparison, the proposed fire detection algorithm works well in various situations, and outperforms the conventional algorithms.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.11
• /
• pp.1647-1654
• /
• 2013

Detection performance of the vehicle on the road depends on weather conditions, the shadow by the movement of the sun, or illumination changes, etc. In this paper, a vehicle detection system in conjunction with a robust background estimate algorithm to environment change on the road in daytime is proposed. Gaussian Mixture Model is applied as background estimation algorithm, and also, Adaboost algorithm is applied to detect the vehicle for candidate region. Through the experiments with input videos obtained from a various weather conditions at the same actual road, the proposed algorithm were useful to detect vehicles in the road.

• International conference on construction engineering and project management
• /
• 2022.06a
• /
• pp.1247-1248
• /
• 2022

Due to rapid urbanization and population growth, it has become crucial to analyze the various volumes and characteristics of pedestrian pathways to understand the capacity and level of service (LOS) for pathways to promote a better walking environment. Different indicators have been developed to measure pedestrian volume. The pedestrian level of service (PLOS), tailored to analyze pedestrian pathways based on the concept of the LOS in transportation in the Highway Capacity Manual, has been widely used. PLOS is a measurement concept used to assess the quality of pedestrian facilities, from grade A (best condition) to grade F (worst condition), based on the flow rate, average speed, occupied space, and other parameters. Since the original PLOS approach has been criticized for producing idealistic results, several modified versions of PLOS have also been developed. One of these modified versions is perceived PLOS, which measures the LOS for pathways by considering pedestrians' awareness levels. However, this method relies on survey-based measurements, making it difficult to continuously deploy the technique to all the pathways. To measure PLOS more quantitatively and continuously, researchers have adopted computer vision technologies to automatically assess pedestrian flows and PLOS from CCTV videos. However, there are drawbacks even with this method because CCTVs cannot be installed everywhere, e.g., in alleyways. Recently, a technique to monitor bio-signals, such as electrodermal activity (EDA), through wearable sensors that can measure physiological responses to external stimuli (e.g., when another pedestrian passes), has gained popularity. It has the potential to continuously measure perceived PLOS. In their previous experiment, the authors of this study found that there were many significant EDA responses in crowded places when other pedestrians acting as external stimuli passed by. Therefore, we hypothesized that the EDA responses would be significantly higher in places where relatively more dynamic objects pass, i.e., in crowded areas with low PLOS levels (e.g., level F). To this end, the authors conducted an experiment to confirm the validity of EDA in inferring the perceived PLOS. The EDA of the subjects was measured and analyzed while watching both the real-world and virtually created videos with different pedestrian volumes in a laboratory environment. The results showed the possibility of inferring the amount of pedestrian volume on the pathways by measuring the physiological reactions of pedestrians. Through further validation, the research outcome is expected to be used for EDA-based continuous measurement of perceived PLOS at the alley level, which will facilitate modifying the existing walking environments, e.g., constructing pathways with appropriate effective width based on pedestrian volume. Future research will examine the validity of the integrated use of EDA and acceleration signals to increase the accuracy of inferring the perceived PLOS by capturing both physiological and behavioral reactions when walking in a crowded area.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.1C
• /
• pp.24-32
• /
• 2010

In this paper we proposed an object tracking system that detects the abandoned and removed objects, which is to be used in the intelligent surveillance applications. After the GMM based background subtraction and by using histogram method, the static region is identified to detect abandoned and removed objects. Since the system is implemented on DSP chip, it operates in realtime and is programmable. The input videos used in the experiment contain various indoor and outdoor scenes, and they are categorized into three different complexities; low, midium and high. By 10 times of experiment, we obtained high detection ratio at low and medium complexity sequences. On the high complexity video, successful detection ratio was relatively low because the scene contains crowdedness and repeated occlusion. In the future work, these complicated situation should be solved.

• Animal Bioscience
• /
• v.36 no.5
• /
• pp.810-817
• /
• 2023

Objective: The aim was to compare the behavioral response to full separation of cows and calves maintained together for 100 days or 24 h. Methods: Twelve Holstein-Friesian cow-calf pairs were enrolled into either treatment or industry groups (n = 6 cow-calf pairs/group). The treatment cows and calves were maintained on pasture together for 106±8.6 d and temporarily separated twice a day for milking. The Industry cows and their calves, were separated within 24 h postpartum. Triaxial accelerometer neck-mounted sensors were fitted to cows 3 weeks before separation to measure hourly rumination and activity. Before separation, cow and calf behavior was observed by scan sampling for 15 min. During the separation process, frequency of vocalizations and turn arounds were recorded. At separation, cows were moved to an observation pen where behavior was recorded for 3 d. A CCTV camera was used to record video footage of cows within the observation pens and behavior was documented from the videos in 15 min intervals across the 3 d. Results: Before separation, industry calves were more likely to be near their mother than Treatment calves. During the separation process, vocalization and turn around behavior was similar between groups. After full separation, treatment cows vocalized three times more than industry cows. However, the frequency of time spent close to barrier, standing, lying, walking, and eating were similar between industry and treatment cows. Treatment cows had greater rumination duration, and were more active, than industry cows. Conclusion: These findings suggest a similar behavioral response to full calf separation and greater occurrence of vocalizations, from cows maintained in a long-term, pasture-based, cow-calf rearing system when ompared to cows separated within 24 h. However, further work is required to assess the impact of full separation on calf behavior.

• Journal of Broadcast Engineering
• /
• v.19 no.6
• /
• pp.798-809
• /
• 2014

Video stabilization is performed to remove unexpected shaky and irregular motion from a video. It is often used as preprocessing for robust feature tracking and matching in video. Typical video stabilization algorithms are developed to compensate motion from surveillance video or outdoor recordings that are captured by a hand-help camera. However, since the vehicle video contains rapid change of motion and local features, typical video stabilization algorithms are hard to be applied as it is. In this paper, we propose a novel approach to compensate shaky and irregular motion in vehicle video using linear regression model and vertical projection histogram matching. Towards this goal, we perform vertical projection histogram matching at each sub region of an input frame, and then we generate linear regression model to extract vertical translation and rotation parameters with estimated regional vertical movement vector. Multiple binarization with sub-region analysis for generating the linear regression model is effective to typical recording environments where occur rapid change of motion and local features. We demonstrated the effectiveness of our approach on blackbox videos and showed that employing the linear regression model achieved robust estimation of motion parameters and generated stabilized video in full automatic manner.