• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.385-387
• /
• 2021

Marine deposition waste threatens the book ecosystem and causes a decrease in catch due to ghost fishing, causing damage of about 370 billion won per year. In order to collect this, a current status survey is conducted using two-way ultrasonic detectors, diving, and lifting frames. However, the scope of the investigation is small to investigate a lot of sedimentary waste, and there is a possibility of causing casualties. This paper deals with the implementation of a high-accuracy marine deposition detection AI model by learning the coastal sediment image data of AI-Hub using the YOLOv5 algorithm suitable for real-time object detection.

• Journal of Korean Society for Geospatial Information Science
• /
• v.21 no.3
• /
• pp.55-62
• /
• 2013

This paper aims to detect Speed Bump by using Omni-directional Camera and to suggest Real-time update scheme of Speed Bump through Vision Based Approach. In order to detect Speed Bump from sequence of camera images, noise should be removed as well as spot estimated as shape and pattern for speed bump should be detected first. Now that speed bump has a regular form of white and yellow area, we extracted speed bump on the road by applying erosion and dilation morphological operations and by using the HSV color model. By collecting huge panoramic images from the camera, we are able to detect the target object and to calculate the distance through GPS log data. Last but not least, we evaluated accuracy of obtained result and detection algorithm by implementing SLAMS (Simultaneous Localization and Mapping system).

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.2
• /
• pp.187-192
• /
• 2021

This study proposes a system for take-off in a forest or similar complex environment using an object detector. In the simulator, a raspberry pi is mounted on a quadcopter with a length of 550mm between motors on a diagonal line, and the experiment is conducted based on edge computing. As for the images to be used for learning, about 150 images of 640⁎480 size were obtained by selecting three points inside Kunsan University, and then converting them to black and white, and pre-processing the binarization by placing a boundary value of 127. After that, we trained the SSD_Inception model. In the simulation, as a result of the experiment of taking off the drone through the model trained with the verification image as an input, a trajectory similar to the takeoff was drawn using the label.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.21 no.1
• /
• pp.105-122
• /
• 2022

According to the statistics about the fatal crashes that have occurred on the expressways for the last 5 years, those who died on the shoulders of the road has been as 3 times high as the others who died on the expressways. It suggests that the crashes on the shoulders of the road should be fatal, and that it would be important to prevent the traffic crashes by cracking down on the vehicles intruding the shoulders of the road. Therefore, this study proposed a method to detect a vehicle that violates the shoulder lane by using the Faster R-CNN. The vehicle was detected based on the Faster R-CNN, and an additional reading module was configured to determine whether there was a shoulder violation. For experiments and evaluations, GTAV, a simulation game that can reproduce situations similar to the real world, was used. 1,800 images of training data and 800 evaluation data were processed and generated, and the performance according to the change of the threshold value was measured in ZFNet and VGG16. As a result, the detection rate of ZFNet was 99.2% based on Threshold 0.8 and VGG16 93.9% based on Threshold 0.7, and the average detection speed for each model was 0.0468 seconds for ZFNet and 0.16 seconds for VGG16, so the detection rate of ZFNet was about 7% higher. The speed was also confirmed to be about 3.4 times faster. These results show that even in a relatively uncomplicated network, it is possible to detect a vehicle that violates the shoulder lane at a high speed without pre-processing the input image. It suggests that this algorithm can be used to detect violations of designated lanes if sufficient training datasets based on actual video data are obtained.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.11
• /
• pp.305-313
• /
• 2019

Recently, spatial information is being constructed actively based on the images obtained by drones. Because occlusion areas occur due to buildings as well as many obstacles, such as trees, pedestrians, and banners in the urban areas, an efficient way to resolve the problem is necessary. Instead of the traditional way, which replaces the occlusion area with other images obtained at different positions, various models based on deep learning were examined and compared. A comparison of a type of feature descriptor, HOG, to the machine learning-based SVM, deep learning-based DNN, CNN, and RNN showed that the CNN is used broadly to detect and classify objects. Until now, many studies have focused on the development and application of models so that it is impossible to select an optimal model. On the other hand, the upgrade of a deep learning-based detection and classification technique is expected because many researchers have attempted to upgrade the accuracy of the model as well as reduce the computation time. In that case, the procedures for generating spatial information will be changed to detect the occlusion area and replace it with simulated images automatically, and the efficiency of time, cost, and workforce will also be improved.

• Journal of IKEEE
• /
• v.27 no.3
• /
• pp.345-349
• /
• 2023

In this paper, we propose a study on the development of deep learning structure for defective pixel detection of next-generation smart LED display board using imaging device. In this research, a technique utilizing imaging devices and deep learning is introduced to automatically detect defects in outdoor LED billboards. Through this approach, the effective management of LED billboards and the resolution of various errors and issues are aimed. The research process consists of three stages. Firstly, the planarized image data of the billboard is processed through calibration to completely remove the background and undergo necessary preprocessing to generate a training dataset. Secondly, the generated dataset is employed to train an object recognition network. This network is composed of a Backbone and a Head. The Backbone employs CSP-Darknet to extract feature maps, while the Head utilizes extracted feature maps as the basis for object detection. Throughout this process, the network is adjusted to align the Confidence score and Intersection over Union (IoU) error, sustaining continuous learning. In the third stage, the created model is employed to automatically detect defective pixels on actual outdoor LED billboards. The proposed method, applied in this paper, yielded results from accredited measurement experiments that achieved 100% detection of defective pixels on real LED billboards. This confirms the improved efficiency in managing and maintaining LED billboards. Such research findings are anticipated to bring about a revolutionary advancement in the management of LED billboards.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.6
• /
• pp.227-233
• /
• 2019

Recently, along with the recent development of deep learning technique, neural networks are achieving success in computer vision filed. Convolutional neural network have shown outstanding performance in not only for a simple image classification task, but also for tasks with high difficulty such as object segmentation and detection. However many such deep learning models are based on supervised-learning, which requires more annotation labels than image-level label. Especially image semantic segmentation model requires pixel-level annotations for training, which is very. To solve these problems, this paper proposes a weakly-supervised semantic segmentation method which requires only image level label to train network. Existing weakly-supervised learning methods have limitations in detecting only specific area of object. In this paper, on the other hand, we use multi-classifier deep learning architecture so that our model recognizes more different parts of objects. The proposed method is evaluated using VOC 2012 validation dataset.

• Journal of Korea Multimedia Society
• /
• v.13 no.3
• /
• pp.417-429
• /
• 2010

Though background subtraction has been widely studied for last decades, it is still a poorly solved problem especially when it meets real environments. In this paper, we first address some common problems for background subtraction that occur in real environments and then those problems are resolved by improving an existing GMM-based background modeling method. First, to achieve low computations, fixed point operations are used. Because background model usually does not require high precision of variables, we can reduce the computation time while maintaining its accuracy by adopting fixed point operations rather than floating point operations. Secondly, to avoid erroneous backgrounds that are induced by high pedestrian traffic, static levels of pixels are examined using shot-time statistics of pixel history. By using a lower learning rate for non-static pixels, we can preserve valid backgrounds even for busy scenes where foregrounds dominate. Finally, to adapt rapid illumination changes, we estimated the intensity change between two consecutive frames as a linear transform and compensated learned background models according to the estimated transform. By applying the fixed point operation to existing GMM-based method, it was able to reduce the computation time to about 30% of the original processing time. Also, experiments on a real video with high pedestrian traffic showed that our proposed method improves the previous background modeling methods by 20% in detection rate and 5~10% in false alarm rate.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.1
• /
• pp.93-101
• /
• 2021

In this paper, a vehicle license plate detection scheme is proposed that uses the spatial attention areas to detect accurately the license plates in various real-road situations. First, the previous WPOD-NET was analyzed, and its detection accuracy is evaluated as lower due to the unnecessary noises in the wide detection candidate areas. To resolve this problem, a vehicle license plate detection model is proposed that uses the candidate area of the license plate as a spatial attention areas. And we compared its performance to that of the WPOD-NET, together with the case of using the optimal spatial attention areas using the ground truth data. The experimental results show that the proposed model has about 20% higher detection accuracy than the original WPOD-NET since the proposed scheme uses tight detection candidate areas.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.10
• /
• pp.1-9
• /
• 2022

In this paper, we propose the Deep Learning-Based Companion Animal Abnormal Behavior Detection Service, which using video and sensor data. Due to the recent increase in households with companion animals, the pet tech industry with artificial intelligence is growing in the existing food and medical-oriented companion animal market. In this study, companion animal behavior was classified and abnormal behavior was detected based on a deep learning model using various data for health management of companion animals through artificial intelligence. Video data and sensor data of companion animals are collected using CCTV and the manufactured pet wearable device, and used as input data for the model. Image data was processed by combining the YOLO(You Only Look Once) model and DeepLabCut for extracting joint coordinates to detect companion animal objects for behavior classification. Also, in order to process sensor data, GAT(Graph Attention Network), which can identify the correlation and characteristics of each sensor, was used.