• 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 Animal Science and Technology
• /
• v.63 no.2
• /
• pp.367-379
• /
• 2021

The objectives of this study were to evaluate convolutional neural network models and computer vision techniques for the classification of swine posture with high accuracy and to use the derived result in the investigation of the effect of dietary fiber level on the behavioral characteristics of the pregnant sow under low and high ambient temperatures during the last stage of gestation. A total of 27 crossbred sows (Yorkshire × Landrace; average body weight, 192.2 ± 4.8 kg) were assigned to three treatments in a randomized complete block design during the last stage of gestation (days 90 to 114). The sows in group 1 were fed a 3% fiber diet under neutral ambient temperature; the sows in group 2 were fed a diet with 3% fiber under high ambient temperature (HT); the sows in group 3 were fed a 6% fiber diet under HT. Eight popular deep learning-based feature extraction frameworks (DenseNet121, DenseNet201, InceptionResNetV2, InceptionV3, MobileNet, VGG16, VGG19, and Xception) used for automatic swine posture classification were selected and compared using the swine posture image dataset that was constructed under real swine farm conditions. The neural network models showed excellent performance on previously unseen data (ability to generalize). The DenseNet121 feature extractor achieved the best performance with 99.83% accuracy, and both DenseNet201 and MobileNet showed an accuracy of 99.77% for the classification of the image dataset. The behavior of sows classified by the DenseNet121 feature extractor showed that the HT in our study reduced (p < 0.05) the standing behavior of sows and also has a tendency to increase (p = 0.082) lying behavior. High dietary fiber treatment tended to increase (p = 0.064) lying and decrease (p < 0.05) the standing behavior of sows, but there was no change in sitting under HT conditions.

• Journal of Korean Society of Transportation
• /
• v.33 no.3
• /
• pp.284-293
• /
• 2015

The objective of this paper is to determine the conditions of road surface by utilizing the images collected from closed-circuit television (CCTV) cameras installed on roadside. First, a technique was examined to detect wet surfaces at nighttime. From the literature reviews, it was revealed that image processing using polarization is one of the preferred options. However, it is hard to use the polarization characteristics of road surface images at nighttime because of irregular or no light situations. In this study, we proposes a new discriminant for detecting wet and dry road surfaces using CCTV image data at night. To detect the road surface conditions with night vision, we applied the wavelet packet transform for analyzing road surface textures. Additionally, to apply the luminance feature of night CCTV images, we set the intensity histogram based on HSI(Hue Saturation Intensity) color model. With a set of 200 images taken from the field, we constructed a detection criteria hyperplane with SVM (Support Vector Machine). We conducted field tests to verify the detection ability of the wet road surfaces and obtained reliable results. The outcome of this study is also expected to be used for monitoring road surfaces to improve safety.

• The Journal of Art Theory & Practice
• /
• no.14
• /
• pp.87-124
• /
• 2012

This paper focuses on the social implication of new media art, which has evolved with the advance of technology. To understand the notion of human-computer interactivity in media art, it examines the meaning of "cybernetics" theory invented by Norbert Wiener just after WWII, who provided "control and communication" as central components of his theory of messages. It goes on to investigate the application of cybernetics theory onto art since the 1960s, to which Roy Ascott made a significant contribution by developing telematic art, utilizing the network of telecommunication. This paper underlines the significance of the relationship between human and machine, art and technology in transforming the work of art as a site of communication and experience. The interactivity in new media art transforms the viewer into the user of the work, who is now provided free will to make decisions on his or her action with the work. The artist is no longer a godlike figure who determines the meaning of the work, yet becomes another user of his or her own work, with which to interact. This paper believes that the interaction between man and machine, art and technology can lead to various ways of interaction between humans, thereby restoring a sense of community while liberating humans from conventional limitations on their creativity. This paper considers the development of new media art more than a mere invention of new aesthetic styles employing advanced technology. Rather, new media art provides a critical shift in subverting the modernist autonomy that advocates the medium specificity. New media art envisions a new art, which would embrace impurity into art, allowing the coexistence of autonomy and heteronomy, embracing a technological other, thereby expanding human relations. By enabling the birth of the user in experiencing the work, interactive new media art produces an open arena, in which the user can create the work while communicating with the work and other users. The user now has freedom to visit the work, to take a journey on his or her own, and to make decisions on what to choose and what to do with the work. This paper contends that there is a significant parallel between new media artists' interest in creating new experiences of the art and Jacques Ranci$\grave{e}$re's concept of the aesthetic regime of art. In his argument for eliminating hierarchy in art and for embracing impurity, Ranci$\grave{e}$re provides a vision for art, which is related to life and ultimately reshapes life. Ranci$\grave{e}$re's critique of both formalist modernism and Jean-Francois Lyotard's postmodern view underlines the social implication of new media art practices, which seek to form "the common of a community."

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.40 no.2
• /
• pp.1-9
• /
• 2003

We develop an experimental system to inspect a metal mask and, in this paper, introduce its inspection algorithm. This system is composed of an ASC(Area Scan Camera) and a belt type xy-table. The whole area of the metal mask is divided into several inspection blocks. The area of each block is equal to FOV(Field of View). For each block, the camera image is compared to the reference image. The reference image is made by gerber file. The rotation angle of the metal mask is calculated through the linear equation that is substituted two end points of horizontal boundary of a specific hole in a camera image. To calculate the position error caused by the belt type xy-table, HT(Hough-Transform) using distances among the holes in two images is used. The center of the reference image is moved as much as the calculated Position error to be coincided with the camera image. The information of holes in each image, such as centroid, size, width and height, are calculated through labeling. Whether a holes is mado correctly by laser machine or not, is judged by comparing the centroid and the size of hole in each image. Finally, we build the experimental system and apply this algorithm.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.27 no.2
• /
• pp.363-368
• /
• 2021

Ship engine rooms have improved automation systems owing to the advancement of technology. However, there are many variables at sea, such as wind, waves, vibration, and equipment aging, which cause loosening, cutting, and leakage, which are not measured by automated systems. There are cases in which only one engineer is available for patrolling. This entails many risk factors in the engine room, where rotating equipment is operating at high temperature and high pressure. When the engineer patrols, he uses his five senses, with particular high dependence on vision. We hereby present a preliminary study to implement an engine-room patrol robot that detects and informs the machine room while a robot patrols the engine room. Images of ship engine-room equipment were classified using a convolutional neural network (CNN). After constructing the image dataset of the ship engine room, the network was trained with a pre-trained CNN model. Classification performance of the trained model showed high reproducibility. Images were visualized with a class activation map. Although it cannot be generalized because the amount of data was limited, it is thought that if the data of each ship were learned through transfer learning, a model suitable for the characteristics of each ship could be constructed with little time and cost expenditure.

• Journal of Biosystems Engineering
• /
• v.30 no.1 s.108
• /
• pp.25-31
• /
• 2005

There are more than three hundred garlic peeling facilities in Korea and most of them use pressurized air for skin peeling operation. One of the major problems of using air for the peeling operation is the occurrence of excessive bruises on the flesh of peeled garlic which causes easy microbial contamination and shortening of the shelf lift. To reduce the occurrence of bruises during the peeling operation, a new type of garlic peeling system was developed which use pressurized water. In this system, high pressure water jets were used to separate garlic bulbs and to peel the skin of garlic cloves. Six commercial systems of this type had been developed and installed at several locations in Korea. The design and performance of the latest system according to three pressure levels were described in this paper. Peeling efficiency of the system was as high as $64.7\%$ in one cycle of peeling operation by three chambers installed in series. Incorporation of a sorting system based on machine vision and re-circulation of unpeeled and partially-peeled garlic enhanced peeling efficiency by additional $30\%$, resulting in total peeling efficiency of the final products of approximately $95\%$. Peeling capacity of the system was over 400 kg per hour.

• The Journal of the Korea Contents Association
• /
• v.9 no.4
• /
• pp.1-11
• /
• 2009

To classifying and filtering of adult images, in recent the computer vision techniques are actively investigated because rapidly increase for the amount of adult images accessible on the Internet. In this paper, we investigate and develop the tool filtering of adult images using skin color model. The tool is consisting of two steps. In the first step, we use a skin color classifier to extract skin color regions from an image. In the nest step, we use a region feature classifier to determine whether an image is an adult image or not an adult image depending on extracted skin color regions. Using histogram color model, a skin color classifier is trained for RGB color values of adult images and not adult images. Using SVM, a region feature classifier is trained for skin color ratio on 29 regions of adult images. Experimental results show that suggested classifier achieve a detection rate of 92.80% with 6.73% false positives.

• Journal of Institute of Control, Robotics and Systems
• /
• v.22 no.2
• /
• pp.139-146
• /
• 2016

Recently, popularity of 3D technology has been growing significantly and it has many application parts in the various fields of industry. In order to overcome the limitations of 2D machine vision technologies based on 2D image, we need the 3D measurement technologies. There are many 3D measurement methods as such scanning probe microscope, phase shifting interferometry, confocal scanning microscope, white-light scanning interferometry, and so on. In this paper, we have used the extended depth of focus (EDF) algorithm among 3D measurement methods. The EDF algorithm is the method which extracts the 3D information from 2D images acquired by short range depth camera. In this paper, we propose the EDF algorithm using the edge informations of images and the average values of all pixel on z-axis to improve the performance of conventional method. To verify the performance of the proposed method, we use the various synthetic images made by point spread function(PSF) algorithm. We can correctly make a comparison between the performance of proposed method and conventional one because the depth information of these synthetic images was known. Through the experimental results, the PSNR of the proposed algorithm was improved about 1 ~ 30 dB than conventional method.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.1
• /
• pp.161-171
• /
• 2014

When the Hough transform is applied to identify an instance of a given model, the output is typically a histogram of votes cast by a set of image features into a parameter space. The next step is to threshold the histogram of counts to hypothesize a given match. The question is "What is a reasonable choice of the threshold?" In a standard implementation of the Hough transform, the threshold is selected heuristically, e.g., some fraction of the highest cell count. Setting the threshold too low can give rise to a false alarm of a given shape(Type I error). On the other hand, setting the threshold too high can result in mis-detection of a given shape(Type II error). In this paper, we derive two conditional probability functions of cell counts in the accumulator array of the generalized Hough transform(GHough), that can be used to select a scientific threshold at the peak detection stage of the Ghough.