• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.2
• /
• pp.631-653
• /
• 2020

Sign language is a natural, visually oriented and non-verbal communication channel between people that facilitates communication through facial/bodily expressions, postures and a set of gestures. It is basically used for communication with people who are deaf or hard of hearing. In order to understand such communication quickly and accurately, the design of a successful sign language translation system is considered in this paper. The proposed system includes object detection and classification stages. Firstly, Single Shot Multi Box Detection (SSD) architecture is utilized for hand detection, then a deep learning structure based on the Inception v3 plus Support Vector Machine (SVM) that combines feature extraction and classification stages is proposed to constructively translate the detected hand gestures. A sign language fingerspelling dataset is used for the design of the proposed model. The obtained results and comparative analysis demonstrate the efficiency of using the proposed hybrid structure in sign language translation.

• Journal of Korea Multimedia Society
• /
• v.18 no.10
• /
• pp.1189-1196
• /
• 2015

In this paper, we proposed automatic extraction of brain tumor using morphological operation and statistical tumors size in MR images. Neurosurgery have used gamma-knife therapy by MR images. However, the gamma-knife plan systems needs the brain tumor regions, because gamma-ray should intensively radiate to the brain tumor except for normal cells. Therefore, gamma-knife plan systems spend too much time on designating the tumor regions. In order to reduce the time of designation of tumors, we progress the automatical extraction of tumors using proposed method. The proposed method consist of two steps. First, the information of skull at MRI slices remove using statistical tumors size. Second, the ROI is extracted by tumor feature and average of tumors size. The detection of tumor is progressed using proposed and threshold method. Moreover, in order to compare the effeminacy of proposed method, we compared snap-shot and results of proposed method.

• The Plant Pathology Journal
• /
• v.35 no.6
• /
• pp.635-643
• /
• 2019

To detect Xanthomonas arboricola pv. pruni, a loopmediated isothermal amplification (LAMP) detection method were developed. The LAMP assay was designed to test crude plant tissue without pre-extraction, or heating incubation, and without advanced analysis equipment. The LAMP primers were designed by targeting an ABC transporter ATP-binding protein, this primer set was tested using the genomic DNA of Xanthomonas and non-Xanthomonas strains, and a ladder product was generated from the genomic DNA of X. arboricola pv. pruni strain but not from 12 other Xanthomonas species strains and 6 strains of other genera. The LAMP conditions were checked with the healthy leaves of 31 peach varieties, and no reaction was detected using either the peach leaves or the peach DNA as a template. Furthermore, the high diagnostic accuracy of the LAMP method was confirmed with 13 X. arboricola pv. pruni strains isolated from various regions in Korea, with all samples exhibiting a positive reaction in LAMP assays. In particular, the LAMP method successfully detected the pathogen in diseased peach leaves and fruit in the field, and the LAMP conditions were proven to be a reliable diagnostic method for the specific detection and identification of X. arboricola pv. pruni in peach orchards.

• Journal of Korea Multimedia Society
• /
• v.10 no.5
• /
• pp.559-568
• /
• 2007

Anchor-person scene detection is of significance for video shot semantic parsing and indexing clues extraction in content-based news video indexing and retrieval system. This paper proposes an efficient algorithm extracting anchor ranges that exist in sports news video for unit structuring of sports news. To detect anchor person scenes, first, anchor person candidate scene is decided by DCT coefficients and motion vector information in the MPEG4 compressed video. Then, from the candidate anchor scenes, image processing method is utilized to classify the news video into anchor-person scenes and non-anchor(sports) scenes. The proposed scheme achieves a mean precision and recall of 98% in the anchor-person scenes detection experiment.

• Journal of IKEEE
• /
• v.23 no.2
• /
• pp.448-453
• /
• 2019

Under the conflicting objectives of maximum user satisfaction and fast launching, there exist great needs for automated mobile app testing. In automated app testing, detection of usage-state changes is one of the most important issues for minimizing human intervention and testing of various usage scenarios. Because conventional approaches utilizing pre-collected training examples can not handle the rapid evolution of apps, we propose a novel method detecting changes in usage-state through graph-entropy. In the proposed method, widgets in a screen shot are recognized through DNNs and 'onverted graphs. We compared the performance of the proposed method with a SIFT (Scale-Invariant Feature Transform) based method on 20 real-world apps. In most cases, our method achieved superior results, but we found some situations where further improvements are required.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.6
• /
• pp.1703-1712
• /
• 1999

In the paper, we propose efficient article and scene change detection algorithms to make the index of sports news compressed in MPEG-2 domain. In the proposed algorithm, the information in MPEG-2 compressed domain is directly used without decoding to save the computation time. The scene change detection algorithm is constructed in an hierarchical method so that the time for detection can be greatly reduced. Also, the algorithm can provide the robust detection against abrupt illuminance change because the luminance and chrominance components are simultaneously considered. Also, the scene change caused by special effect such as dissolve and wipe can be detected in the compressed domain. In the article detection, the algorithm is constructed for robust detection of the anchor frame using the concept of CCV(Color Coherent Vector).

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.12
• /
• pp.3904-3922
• /
• 2022

As a research hotspot, pedestrian detection has a wide range of applications in the field of computer vision in recent years. However, current pedestrian detection methods have problems such as insufficient detection accuracy and large models that are not suitable for large-scale deployment. In view of these problems mentioned above, a lightweight pedestrian detection and early warning method using a new model called you only look once (Yolov5) is proposed in this paper, which utilizing advantages of Yolov5s model to achieve accurate and fast pedestrian recognition. In addition, this paper also optimizes the loss function of the batch normalization (BN) layer. After sparsification, pruning and fine-tuning, got a lot of optimization, the size of the model on the edge of the computing power is lower equipment can be deployed. Finally, from the experimental data presented in this paper, under the training of the road pedestrian dataset that we collected and processed independently, the Yolov5s model has certain advantages in terms of precision and other indicators compared with traditional single shot multiBox detector (SSD) model and fast region-convolutional neural network (Fast R-CNN) model. After pruning and lightweight, the size of training model is greatly reduced without a significant reduction in accuracy, and the final precision reaches 87%, while the model size is reduced to 7,723 KB.

• Korean Journal of Poultry Science
• /
• v.37 no.2
• /
• pp.181-190
• /
• 2010

Mycoplasma gallisepticum (MG) infection results in severe economic loss in the poultry industry. In the previous reports, F strain, one of the MG live vaccine strains, could protect the bird from field infection of MG strains. In this study, efficacy of PoulShot$^{(R)}$ MG-F vaccine againset mycoplasma gallisepticum infection was evaluated for filed application in commercial layers. Commercial layers from two different farms received with PoulShot$^{(R)}$ MG-F, MG-F live vaccine at 9~14 weeks of age. Serological immune response to MG vaccine, the persistency of MG vaccine strain in the upper respiratory tracts and egg production rate were evaluated in the vaccinated, contacted or nonvaccinated flocks. The serological response was first detected at 3 weeks after vaccination (WAV) and persisted for 31 WAV. The MG vaccine strains were also persisted for 31 WAV based on the reisolation and PCR detection. There was no difference between the vaccinated or non-vaccinated flocks in the egg production rate but in the abnormality rate of eggs. Based on the above results, we suggested that the PoulShot$^{(R)}$, MG-F live vaccine was fully immunogenic and had characteristics of long persistence in the upper respiratory trachea which will reduce economic loss caused by MG infection in the layer farms.

• The KIPS Transactions:PartB
• /
• v.9B no.3
• /
• pp.343-350
• /
• 2002

A scene change detection which involves finding a cut between two consecutive shots is an important step for video indexing and retrieval. This paper proposes an algorithm for fast and accurate detection of abrupt scene changes in an MPEG compressed domain with minimal decoding requirements arid computational effort. The proposed method compares two successive DC images of I-frames for finding the GOP (group of picture) which contain a scene change and uses macroblock-coded type information contained in B-frames to detect the exact frame where the scene change occurred. The experiment results demonstrate that the proposed algorithm has better detection performance, such as precision and recall rate, than the existing method using all DC images. The algorithm has the advantage of speed, simplicity and accuracy. In addition, it requires less amount of storage.

• 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.