• Journal of the Korean Society of Clothing and Textiles
• /
• v.30 no.3 s.151
• /
• pp.412-424
• /
• 2006

This paper starts out with a review of the concepts of naturalism and natural clothes. Based on these concepts, the paper aims to create nature-friendly and high-value modem designs by grafting natural images from butterflies into human bodies. The images of butterflies are combined to show distinguished luxury and diversity with natural materials such as denim. As for the design technology, Adove Illustrator 10, Photoshop 7.0, and Prima Vision Textile Design System are used since they are very popular in the CAD system. The followings are the main findings of our research. Naturalism displays its objectivity based on its inherent interest in human nature and its factual description in scientific approaches with an intention to find a true meaning of human life in nature. Natural clothes can be defined as the clothes of natural silhouette which are made of pure materials in natural colors and dyes from natural motifs such as animals, plants or natural phenomena. We have realized that the visually distinguished looks of butterflies are indeed far more luxurious than those of any other insects and this visual distinction can be utilized to attract the attention from designers. Lastly, it turns out that the digital printing technique can create more variety in colors and design forms than the manual printing, thereby providing designers with more options and practicality.

• 재활복지
• /
• v.21 no.3
• /
• pp.129-146
• /
• 2017

In an event of an emergency, such as fire in a building, visually impaired and blind people are prone to exposed to a level of danger that is greater than that of normal people, for they cannot be aware of it quickly. Current fire detection methods such as smoke detector is very slow and unreliable because it usually uses chemical sensor based technology to detect fire particles. But by using vision sensor instead, fire can be proven to be detected much faster as we show in our experiments. Previous studies have applied various image processing and machine learning techniques to detect fire, but they usually don't work very well because these techniques require hand-crafted features that do not generalize well to various scenarios. But with the help of recent advancement in the field of deep learning, this research can be conducted to help solve this problem by using deep learning-based object detector that can detect fire using images from security camera. Deep learning based approach can learn features automatically so they can usually generalize well to various scenes. In order to ensure maximum capacity, we applied the latest technologies in the field of computer vision such as YOLO detector in order to solve this task. Considering the trade-off between recall vs. complexity, we introduced two convolutional neural networks with slightly different model's complexity to detect fire at different recall rate. Both models can detect fire at 99% average precision, but one model has 76% recall at 30 FPS while another has 61% recall at 50 FPS. We also compare our model memory consumption with each other and show our models robustness by testing on various real-world scenarios.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.3
• /
• pp.429-438
• /
• 2021

Owing to the development of science technology, particularly the smart concept and defense policy factors of the 4th industry, military weapon systems are advanced, and the scientific and operational force is reduced dramatically. The aspect of the future war is characterized by the operation of troops with reduced forces from advanced and scientific weapon systems in an operational area that has expanded more than four times compared to the present. Reflecting on these situational factors, it is necessary to improve combat methods based on the changes in the battlefield environment and advanced weapon systems. In this study, to find a more efficient future combat method in a changing war pattern, this study applied the battle experiment methodology using Vision21 war game model, which is an analytical model used by the army. Finally, this study aimed to verify the future combat method and unit structure. Therefore, the scenario composition and experiment method that reflect the change in the ground operational environment and weapon system was first composed. Subsequently, an analysis method based on the combat effectiveness was applied to verify the effective combat performance method and unit structure of future infantry units.

• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.1-16
• /
• 2022

Despite recent breakthroughs in deep learning and computer vision fields, the pixel-wise identification of tiny objects in high-resolution images with complex disturbances remains challenging. This study proposes a modified U-net for tiny crack segmentation in real-world steel-box-girder bridges. The modified U-net adopts the common U-net framework and a novel Self-Attention-Self-Adaption (SASA) neuron as the fundamental computing element. The Self-Attention module applies softmax and gate operations to obtain the attention vector. It enables the neuron to focus on the most significant receptive fields when processing large-scale feature maps. The Self-Adaption module consists of a multiplayer perceptron subnet and achieves deeper feature extraction inside a single neuron. For data augmentation, a grid-based crack random elastic deformation (CRED) algorithm is designed to enrich the diversities and irregular shapes of distributed cracks. Grid-based uniform control nodes are first set on both input images and binary labels, random offsets are then employed on these control nodes, and bilinear interpolation is performed for the rest pixels. The proposed SASA neuron and CRED algorithm are simultaneously deployed to train the modified U-net. 200 raw images with a high resolution of 4928 × 3264 are collected, 160 for training and the rest 40 for the test. 512 × 512 patches are generated from the original images by a sliding window with an overlap of 256 as inputs. Results show that the average IoU between the recognized and ground-truth cracks reaches 0.409, which is 29.8% higher than the regular U-net. A five-fold cross-validation study is performed to verify that the proposed method is robust to different training and test images. Ablation experiments further demonstrate the effectiveness of the proposed SASA neuron and CRED algorithm. Promotions of the average IoU individually utilizing the SASA and CRED module add up to the final promotion of the full model, indicating that the SASA and CRED modules contribute to the different stages of model and data in the training process.

• Journal of Practical Engineering Education
• /
• v.11 no.2
• /
• pp.175-180
• /
• 2019

Existing fashion web sites show only the search results for one type of clothes in items such as tops and bottoms. As the fashion market grows, consumers are demanding a platform to find a variety of fashion information. To solve this problem, we devised the idea of linking image classification through deep learning with a website and integrating SNS functions. User uploads their own image to the web site and uses the deep learning server to identify, classify and store the image's characteristics. Users can use the stored information to search for the images in various combinations. In addition, communication between users can be actively performed through the SNS function. Through this, the plan to solve the problem of existing fashion-related sites was prepared.

• Journal of Broadcast Engineering
• /
• v.25 no.1
• /
• pp.67-82
• /
• 2020

This study is based on the screen composition(shot size, split screen, candidate placement, lighting, stage) that fits the discussion process (round table, sitting, standing method) adopted from the 19th presidential candidates' TV joint discussion to analyze the adequacy and limitations of the methodology(design, candidate background, etc.). In addition, the progress and screen composition of the 45th US Presidential Candidates Debate in 2016 and the 25th Presidential Candidates of France TV Joint Debate in 2017 have been analyzed to compare and contrast the methods used in Korea. Through this, we apply the screening method to the 20th Presidential Candidate TV Joint Debate, which will be held in 2022, to fit the fixed discussion process that can fully identify candidates' qualities, policies, and vision.

• Journal of Advanced Information Technology and Convergence
• /
• v.10 no.1
• /
• pp.71-83
• /
• 2020

In this paper, we propose a system that can detect the shape of a hand at high speed using an FPGA. The hand-shape detection system is designed using Verilog HDL, a hardware language that can process in parallel instead of sequentially running C++ because real-time processing is important. There are several methods for hand gesture recognition, but the image processing method is used. Since the human eye is sensitive to brightness, the YCbCr color model was selected among various color expression methods to obtain a result that is less affected by lighting. For the CbCr elements, only the components corresponding to the skin color are filtered out from the input image by utilizing the restriction conditions. In order to increase the speed of object recognition, a median filter that removes noise present in the input image is used, and this filter is designed to allow comparison of values and extraction of intermediate values at the same time to reduce the amount of computation. For parallel processing, it is designed to locate the centerline of the hand during scanning and sorting the stored data. The line with the highest count is selected as the center line of the hand, and the size of the hand is determined based on the count, and the hand and arm parts are separated. The designed hardware circuit satisfied the target operating frequency and the number of gates.

• Journal of the Institute of Convergence Signal Processing
• /
• v.12 no.4
• /
• pp.258-266
• /
• 2011

AR(Augmented Reality) technology is now easily shown around us with respect to its applicable areas' being spreaded into various shapes since the usage is simply generalized and many-sided. Currently existing camera vision based AR used marker based methods rather than using real world's informations. For the marker based AR technology, there are limitations on applicable areas and its environmental properties that a user could immerse into the usage of application program. In this paper, we proposed a novel AR method which users could recognize objects from the real world's data and the related 3-dimensional contents are also displayed. Those are done using image processing skills and a smart mobile embedded camera for terminal based AR implementations without any markers. Object recognition is done from the comparison of pre-registered and referenced images. In this process, we tried to minimize the amount of computations of similarity measurements for improving working speed by considering features of smart mobile devices. Additionally, the proposed method is designed to perform reciprocal interactions through touch events using smart mobile devices after the 3-dimensional contents are displayed on the screen. Since then, a user is able to acquire object related informations through a web browser with respect to the user's choice. With the system described in this paper, we analyzed and compared a degree of object recognition, working speed, recognition error for functional differences to the existing AR technologies. The experimental results are presented and verified in smart mobile environments to be considered as an alternate and appropriate AR technology.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.1
• /
• pp.39-47
• /
• 2023

In this paper, we propose a method to measure the similarity between golf swings in videos. As it is known that deep learning-based artificial intelligence technology is effective in the field of computer vision, attempts to utilize artificial intelligence in video-based sports data analysis are increasing. In this study, the joint coordinates of a person in a golf swing video were obtained using a deep learning-based pose estimation model, and based on this, the similarity of each swing segment was measured. For the evaluation of the proposed method, driver swing videos from the GolfDB dataset were used. As a result of measuring swing similarity by pairing swing videos of a total of 36 players, 26 players evaluated that their other swing sequence was the most similar, and the average ranking of similarity was confirmed to be about 5th. This ensured that the similarity could be measured in detail even when the motion was performed similarly.

• Smart Structures and Systems
• /
• v.31 no.4
• /
• pp.365-381
• /
• 2023

The existing vision-based techniques for inspection and condition assessment of civil infrastructure are mostly manual and consequently time-consuming, expensive, subjective, and risky. As a viable alternative, researchers in the past resorted to deep learning-based autonomous damage detection algorithms for expedited post-disaster reconnaissance of structures. Although a number of automatic damage detection algorithms have been proposed, the scarcity of labeled training data remains a major concern. To address this issue, this study proposed a semi-supervised learning (SSL) framework based on consistency regularization and cross-supervision. Image data from post-earthquake reconnaissance, that contains cracks, spalling, and exposed rebars are used to evaluate the proposed solution. Experiments are carried out under different data partition protocols, and it is shown that the proposed SSL method can make use of unlabeled images to enhance the segmentation performance when limited amount of ground truth labels are provided. This study also proposes DeepLab-AASPP and modified versions of U-Net++ based on channel-wise attention mechanism to better segment the components and damage areas from images of reinforced concrete buildings. The channel-wise attention mechanism can effectively improve the performance of the network by dynamically scaling the feature maps so that the networks can focus on more informative feature maps in the concatenation layer. The proposed DeepLab-AASPP achieves the best performance on component segmentation and damage state segmentation tasks with mIoU scores of 0.9850 and 0.7032, respectively. For crack, spalling, and rebar segmentation tasks, modified U-Net++ obtains the best performance with Igou scores (excluding the background pixels) of 0.5449, 0.9375, and 0.5018, respectively. The proposed architectures win the second place in IC-SHM2021 competition in all five tasks of Project 2.