• Journal of Korea Multimedia Society
• /
• v.24 no.2
• /
• pp.305-311
• /
• 2021

One of the highly reliable object tracking methods is to trace objects by associating objects detected by deep learning. The detected object is represented by a rectangular box. The box has information such as location and size. Since the tracker has motion information of the object in addition to the location and size, knowing additional information about the motion of the detected box can increase the reliability of object tracking. In this paper, we present a new method of reliably estimating the moving direction of the detected object in underground parking lot. First, the frame difference image is binarized for detecting motion energy, change due to the object motion. Then, a cumulative binary image is generated that shows how the motion energy changes over time. Next, the moving direction of the detected box is estimated from the accumulated image. We use a new cost function to accurately estimate the direction of movement of the detected box. The proposed method proves its performance through comparative experiments of the existing methods.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.1
• /
• pp.93-101
• /
• 2023

In this paper, we propose a method of creating a 3D bounding box for an object using a vanishing point to increase the accuracy of object recognition in an image when recognizing an traffic object using a video camera. Recently, when vehicles captured by a traffic video camera is to be detected using artificial intelligence, this 3D bounding box generation algorithm is applied. The vertical vanishing point (VP1) and horizontal vanishing point (VP2) are derived by analyzing the camera installation angle and the direction of the image captured by the camera, and based on this, the moving object in the video subject to analysis is specified. If this algorithm is applied, it is easy to detect object information such as the location, type, and size of the detected object, and when applied to a moving type such as a car, it is tracked to determine the location, coordinates, movement speed, and direction of each object by tracking it. Able to know. As a result of application to actual roads, tracking improved by 10%, in particular, the recognition rate and tracking of shaded areas (extremely small vehicle parts hidden by large cars) improved by 100%, and traffic data analysis accuracy was improved.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.05a
• /
• pp.363-366
• /
• 2016

In this paper, we propose a method for detecting pedestrian, problem-solving situations that occur in a cross. When a pedestrian crossing and other, there occurs a problem of detecting the other pedestrians for detecting a specific pedestrian in the image. The proposed method for solving the problem is as follows. First, select a specific pedestrian detected by bounding box, and extracts the area as a template. Detecting a pedestrian from the image using the HOG, and designated as a candidate region. The final choice of the pedestrian detected by comparison with a candidate pedestrian with the specific pedestrian extracted for template. In comparison, using the Template matching, Histogram comparison and LBP.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.11
• /
• pp.41-48
• /
• 2024

This paper propose an automatic labeling system for fashion items in images by combining one of the object detection models, YOLO(You Only Look Once), with a high-level classification object detection model. After detecting the primary fashion items, TOP and BOTTOM, in an image, the system analysis the bounding boxes of the detected objects and removes redundant or unnecessary bounding boxes through preprocessing to extract bounding boxes with accurate location information. The extracted bounding boxes are compared to the classes defined by the high-level object detection model with coordinate normalization to perform automatic labeling by matcing the input fashion item types. The system's performance was evaluated on 10,000 fashion images and corresponding test data, and 8,192 images were found to be correctly labeled. This demonstrates a significant improvement in efficiency over manual labeling methods, showing the system's practical contribution to large-scale fashion image data processing.

• Journal of Korea Multimedia Society
• /
• v.4 no.5
• /
• pp.396-405
• /
• 2001

Usually it is difficult to extract facial regions in a complex image by using only a predetermined skin color. Expecially, it is more difficult to separate them from background regions that contains the skin color. This paper proposes a face detection method by using gradual range expansion of an initial skin color. By analyzing the skin color distribution several images that are collected in the Web, the range of dense distribution is selected as the range of the initial skin color. In each expanding step, expanded regions in the image are tested whether they can be actual facial regions by using the information of the shape of general face and the location of face organs. The shape of general face is modeled as an ellipse and the aspect ratio of its bounding box is used to define the shape constraint for faces. Only the eyes and lips are used as the face organs, which can be easily detected by extracting horizontal edges in the expanded regions. through several experiments, it is confirmed that the proposed method can detect exactly not only faces having partly distorted regions by highlight but also faces neighboring similar color regions.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.21 no.3
• /
• pp.83-95
• /
• 2022

This paper proposes a system that performs object detection and distance estimation for application to autonomous vehicles. Object detection is performed by a network that adjusts the split grid to the input image ratio using the characteristics of the recently actively used deep learning model YOLOv4, and is trained to a custom dataset. The distance to the detected object is estimated using a bounding box and homography. As a result of the experiment, the proposed method improved in overall detection performance and processing speed close to real-time. Compared to the existing YOLOv4, the total mAP of the proposed method increased by 4.03%. The accuracy of object recognition such as pedestrians, vehicles, construction sites, and PE drums, which frequently occur when driving to the city center, has been improved. The processing speed is approximately 55 FPS. The average of the distance estimation error was 5.25m in the X coordinate and 0.97m in the Y coordinate.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.3
• /
• pp.523-530
• /
• 2024

Online, such as web pages and digital documents, have the ability to search for specific words or specific phrases that users want to search in real time. Printed materials such as printed books and reference books often have difficulty finding specific words or specific phrases in real time. This paper describes the development of a deep learning model for detecting text and a real-time character detection system using OCR for recognizing text. This study proposes a method of detecting text using the EAST model, a method of recognizing the detected text using EasyOCR, and a method of expressing the recognized text as a bounding box by comparing a specific word or specific phrase that the user wants to search for. Through this system, users expect to find specific words or phrases they want to search in real time in print, such as books and reference books, and find necessary information easily and quickly.

• Proceedings of the KIEE Conference
• /
• 2004.11c
• /
• pp.291-294
• /
• 2004

Semantic video object extraction is important for tracking meaningful objects in video and object-based video coding. We propose a fast semiautomatic video object extraction algorithm which combines a watershed segmentation schemes and chamfer distance transform. Initial object boundaries in the first frame are defined by a human before the tracking, and fast video object tracking can be achieved by tracking only motion-detected regions in a video frame. Experimental results shows that the boundaries of tracking video object arc close to real video object boundaries and the proposed algorithm is promising in terms of speed.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.10 no.5
• /
• pp.229-235
• /
• 2010

In this paper, we propose a method to improve the efficiency of the moving object detection in real-time surveillance camera system. The existing methods, the methods using differential image and background image, are difficult to detect the moving object from outside the video streams. The proposed method keeps the background image if it doesn't be detected moving object using the differential value between a previous frame and a current frame. And the background image is renewed as the moving object is gone in a frame. To decide people and wild pig, the proposed system estimates a bounding box enclosing each moving object in the detecting region. As a result of simulation, the proposed method is better than the existing method.

• International journal of advanced smart convergence
• /
• v.10 no.4
• /
• pp.110-116
• /
• 2021

This paper proposed a real-time earlobe detection system using deep learning on the web. Existing deep learning-based detection methods often find independent objects such as cars, mugs, cats, and people. We proposed a way to receive an image through the camera of the user device in a web environment and detect the earlobe on the server. First, we took a picture of the user's face with the user's device camera on the web so that the user's ears were visible. After that, we sent the photographed user's face to the server to find the earlobe. Based on the detected results, we printed an earring model on the user's earlobe on the web. We trained an existing YOLO v5 model using a dataset of about 200 that created a bounding box on the earlobe. We estimated the position of the earlobe through a trained deep learning model. Through this process, we proposed a real-time earlobe detection system on the web. The proposed method showed the performance of detecting earlobes in real-time and loading 3D models from the web in real-time.