• Journal of Intelligence and Information Systems
• /
• v.28 no.4
• /
• pp.329-346
• /
• 2022

Recently, as the market for short form videos (Instagram, TikTok, YouTube) on social media has gradually increased, research using them is actively being conducted in the artificial intelligence field. A representative research field is Video to Shop, which detects fashion products in videos and searches for product images. In such a video-based artificial intelligence model, product features are extracted using convolution operations. However, due to the limitation of computational resources, extracting features using all the frames in the video is practically impossible. For this reason, existing studies have improved the model's performance by sampling only a part of the entire frame or developing a sampling method using the subject's characteristics. In the existing Video to Shop study, when sampling frames, some frames are randomly sampled or sampled at even intervals. However, this sampling method degrades the performance of the fashion product search model while sampling noise frames where the product does not exist. Therefore, this paper proposes a sampling method MF (Missing Fashion items on frame) sampler that removes noise frames and improves the performance of the search model. MF sampler has improved the problem of resource limitations by developing a keyframe mechanism. In addition, the performance of the search model is improved through noise frame removal using the noise detection model. As a result of the experiment, it was confirmed that the proposed method improves the model's performance and helps the model training to be effective.

• Journal of the Society of Disaster Information
• /
• v.18 no.4
• /
• pp.922-929
• /
• 2022

Purpose: The purpose of this paper is to develop a deep learning model to detect fire extinguishers in images taken from CCTVs in underground utility tunnels. Method: Various fire extinguisher images were collected for detection of fire extinguishers in the running-based underground utility tunnel, and a model applying the One-stage Detector method was developed based on the CNN algorithm. Result: The detection rate of fire extinguishers photographed within 10m through CCTV video in the underground common area is over 96%, showing excellent detection rate. However, it was confirmed that the fire extinguisher object detection rate drops sharply at a distance of 10m or more, in a state where it is difficult to see with the naked eye. Conclusion: This paper develops a model for detecting fire extinguisher objects in underground common areas, and the model shows high performance, and it is judged that it can be used for underground common area digital twin model synchronizing.

• The Korean Journal of Applied Statistics
• /
• v.36 no.2
• /
• pp.101-114
• /
• 2023

The data augmentation technique is effectively used to solve the problem of overfitting the model by allowing the training dataset to be viewed from various perspectives. In addition to image augmentation techniques such as rotation, cropping, horizontal flip, and vertical flip, occlusion-based data augmentation methods such as Cutmix and Cutout have been proposed. For models based on speech data, it is possible to use an occlusion-based data-based augmentation technique after converting a 1D speech signal into a 2D spectrogram. In particular, SpecAugment is an occlusion-based augmentation technique for speech spectrograms. In this study, we intend to compare and study data augmentation techniques that can be used in the problem of false-voice detection. Using data from the ASVspoof2017 and ASVspoof2019 competitions held to detect fake audio, a dataset applied with Cutout, Cutmix, and SpecAugment, an occlusion-based data augmentation method, was trained through an LCNN model. All three augmentation techniques, Cutout, Cutmix, and SpecAugment, generally improved the performance of the model. In ASVspoof2017, Cutmix, in ASVspoof2019 LA, Mixup, and in ASVspoof2019 PA, SpecAugment showed the best performance. In addition, increasing the number of masks for SpecAugment helps to improve performance. In conclusion, it is understood that the appropriate augmentation technique differs depending on the situation and data.

• Convergence Security Journal
• /
• v.22 no.3
• /
• pp.25-32
• /
• 2022

Recently, the image processing industry has been activated as deep learning-based technology is introduced in the image recognition and detection field. With the development of deep learning technology, learning model vulnerabilities for adversarial attacks continue to be reported. However, studies on countermeasures against poisoning attacks that inject malicious data during learning are insufficient. The conventional countermeasure against poisoning attacks has a limitation in that it is necessary to perform a separate detection and removal operation by examining the training data each time. Therefore, in this paper, we propose a technique for reducing the attack success rate by applying modifications to the training data and inference data without a separate detection and removal process for the poison data. The One-shot kill poison attack, a clean label poison attack proposed in previous studies, was used as an attack model. The attack performance was confirmed by dividing it into a general attacker and an intelligent attacker according to the attacker's attack strategy. According to the experimental results, when the proposed defense mechanism is applied, the attack success rate can be reduced by up to 65% compared to the conventional method.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.10 no.6
• /
• pp.23-29
• /
• 2010

Nowadays it is required to design intelligent visual surveillance systems which automatically detect abandoned objects in public places to strengthen the social safety. Already recognized abandoned objects can be occluded partially or fully by surrounding people in public places after the first recognition. To improve an essential recognition performance index PAT, the system should overcome the occlusion problems. In this research, a design scheme is newly proposed to construct the robust detection system which is comprised of multiple stages considering the occlusion problem. To show the feasibilities of the proposed system, the evaluation was tried for the prepared image streams including 6 various situations and the experimental results show 96% and 75% in PAT performance for intrusion and abandoning events, respectively. Finally in spite of full occlusions by multiple persons, the proposed system shows the capability to continuously recognize the abandoned object after complex occlusions disappear.

• Spatial Information Research
• /
• v.11 no.1
• /
• pp.13-22
• /
• 2003

The study of geomorphology of tidal flat using remote sensing image has been considered useful because of it's ability to acquire data periodically. Especially, the Near Infrared band of satellite image has been used to divide between land and sea area. This study extracted a borderline of the tidal flat using Landsat-5 images and generated DEM(Digital elevation model) using tide level data as elevation value. DEM is a useful tool for three-dimensional survey of geomorphology and can be used for survey of tidal flat. This study divided 8 images of 1990's into two parts - before 1994 and after 1994 - and generated DEM respectively. In this work, the areas of tidal flats are calculated and it was revealed the area of tidal flat was decreased after 1994.

• Journal of Korea Multimedia Society
• /
• v.22 no.12
• /
• pp.1447-1456
• /
• 2019

As artificial intelligence technology is actively used in image processing, it is possible to generate high-quality fake images based on deep learning. Fake images generated using GAN(Generative Adversarial Network), one of unsupervised learning algorithms, have reached levels that are hard to discriminate from the naked eye. Detecting these fake images is required as they can be abused for crimes such as illegal content production, identity fraud and defamation. In this paper, we develop a deep-learning model based on CNN(Convolutional Neural Network) for the detection of StyleGAN fake images. StyleGAN is one of GAN algorithms and has an excellent performance in generating face images. We experiment with 48 number of experimental scenarios developed by combining parameters of the proposed model. We train and test each scenario with 300,000 number of real and fake face images in order to present a model parameter that improves performance in the detection of fake faces.

• Journal of Advanced Navigation Technology
• /
• v.12 no.2
• /
• pp.164-169
• /
• 2008

This paper is using motion tracking by image segmentation to monitor intruders and to confirm based on mobile the relevant information. First, detect frame in animation that film fixed area, and make use of image subtraction between two frame that adjoin, segment fixed backing and target who move. Segmental foreground object to the edge detecting the location specified by the edge of the median estimate extracted by analyzing the motion of the intruders to monitor. When a motion is detected, the detected image is transmitted by using the W AP pull basis image transmission method on the mobile client data terminal.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.9 no.4
• /
• pp.69-74
• /
• 2009

Abandoned objects should be treated as possibly dangerous things for public areas until they turn out to be safe because explosive material or chemical substance is intentionally contained in them for public terrors. For large public areas such as airports or train stations, there are limits in man-power for security staffs to check all the monitors for covering the entire area under surveillance. This is the basic motivation of developing the automatic detection system for dangerous abandoned objects based on vision technology. In this research, well-known DBE is applied to stably extract background images and the HOG algorithm is adapted to discriminate between human and stuff for object classification. To show the effectiveness of the proposed system, experiments are carried out in detecting intrusion for a forbidden area and alarming for abandoned objects in a room under surveillance.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.10a
• /
• pp.1341-1344
• /
• 2015

In this paper, the designed fish robot is researched and developed for aquarium underwater robot. This paper is a study on how the outside technology merely to find the location of fish robots without specific sensor or internal devices. This model is designed to detect the position of the Robotic Fish in the Mat lab and Simulink. This intends to recognize the shape of the tank via a video device such as a camera or camcorder using an image processing technique to identify the location of the robotic fishes. Here, we are applied the two methods, one is Hom - Schunk Method and second one is newly proposed method that is the comparing image data algorithm. The Horn - Schunck Method is used to obtain the velocity for each pixel in the image and the comparing image data algorithm is proposed to obtain the position with comparing two video frames and assumes a constant velocity in each video frame.