A study on agricultural automation became more important. In Korea, sweet persimmon farmers spend a lot of time and effort on classifying profitable persimmons. In this paper, we propose and implement an efficient grading algorithm for persimmons before shipment. We gathered more than 1,750 images of persimmons, and the images were graded and labeled for classifications purpose. Our main algorithm is based on EfficientDet object detection model but we implemented more exquisite method for better classification performance. In order to improve the precision of classification, we adopted a machine learning algorithm, which was proposed by PyCaret machine learning workflow generation library. Finally we acquired an improved classification model with the accuracy score of 81%.

Versatile Video Coding (VVC) is the latest video coding standard, which had been developed by the Joint Video Experts Team (JVET) of ITU-T Video Coding Experts Group (VCEG) and ISO/IEC Moving Picture Experts Group (MPEG) in 2020. Although VVC can provide powerful coding performance, it requires tremendous computational complexity to determine the optimal block structures during the encoding process. In this paper, we propose a fast ternary tree decision method using two neural networks with 7 nodes as input vector based on the multi-layer perceptron structure, names STH-NN and STV-NN. As a training result of neural network, the STH-NN and STV-NN achieved accuracies of 85% and 91%, respectively. Experimental results show that the proposed method reduces the encoding complexity up to 25% with unnoticeable coding loss compared to the VVC test model (VTM).

As the IMO introduced MASS (Maritime Autonomous Surface Ships), ISO(International Organization for Standardization) announced ISO 19847 of a maritime data sharing standard for collecting and remotely managing data of ship systems. Previous literature evaluated the risk using HAZOP for ISO 19847 and proved that risk assessment is useful through experiments. However, redundancy of ISO 19847 ship data server which is one of the risk reduction method suggested in previous literature, was designed but couldn't tested due to the limitations of the conditions. So, in this study, to prove the usefulness of the ship data server redundancy of ISO 19847 which was not tested in previous literature. It based on the design of previous literature, and the network of ship data servers was modeled using the SES/DEVS format and simulated using the DEVS# open source library.

In order to further improve the accuracy and time efficiency of behavior recognition in intelligent monitoring scenarios, a human behavior recognition algorithm based on YOLO combined with LSTM and CNN is proposed. Using the real-time nature of YOLO target detection, firstly, the specific behavior in the surveillance video is detected in real time, and the depth feature extraction is performed after obtaining the target size, location and other information; Then, remove noise data from irrelevant areas in the image; Finally, combined with LSTM modeling and processing time series, the final behavior discrimination is made for the behavior action sequence in the surveillance video. Experiments in the MSR and KTH datasets show that the average recognition rate of each behavior reaches 98.42% and 96.6%, and the average recognition speed reaches 210ms and 220ms. The method in this paper has a good effect on the intelligence behavior recognition.

Cross-domain recommender system transfers knowledge across different domains to improve the recommendation performance in a target domain that has a relatively sparse model. However, they suffer from the "negative transfer" in which transferred knowledge operates as noise. This paper proposes a novel Multiple Fusion-based Deep Cross-Domain Recommendation named MFDCR. We exploit Doc2Vec, one of the famous word embedding techniques, to fuse data user-wise and transfer knowledge across multi-domains. It alleviates the "negative transfer" problem. Additionally, we introduce a simple multi-layer perception to learn the user-item interactions and predict the possibility of preferring items by users. Extensive experiments with three domain datasets from one of the most famous services Amazon demonstrate that MFDCR outperforms recent single and cross-domain recommendation algorithms. Furthermore, experimental results show that MFDCR can address the problem of "negative transfer" and improve recommendation performance for multiple domains simultaneously. In addition, we show that our approach is efficient in extending toward more domains.

Based on the problems of untimely Expressway fire rescue and backward traditional fire rescue methods, an emergency fire fighting and inspection robot riding on expressway guardrail is designed. The overall mechanical structure design of emergency fire fighting and inspection robot riding on expressway guardrail is completed by using three-dimensional design software. The target fire detection is realized by using the target detection algorithm of Yolov5; By selecting a variety of sensors and using the control method of multi algorithm fusion, the basic function of robot on duty early warning is realized, and it has the ability of intelligent fire extinguishing. The BMS battery charging and discharging system is used to detect the real-time power of the robot. The design of the expressway emergency fire fighting and inspection robot provides a new technical means for the development of emergency fire fighting equipment, and improves the reliability and efficiency of expressway emergency fire fighting.

This study intends to propose an evaluation tool for analyzing the sense of immersion of realistic VR game contents in the metabus game industry, which is rapidly increasing recently. This paper presents an analysis tool to measure the degree of immersion after selecting three types VR games for research, and analyzing the immersion way method of the game. The research method analyzes how the user and the sense of immersion formed in the storytelling in the game can have an effect. Based on the results, we present an sense of immersion analysis and measurement method for evaluating the usability of VR game contents, which is a key element in the metaverse era. Furthermore, it expands the immersion element that appears in storytelling to the game industry and seeks a new direction of development through connection with the VR game industry.

During the COVID-19 pandemic, games played an important role in people's lives. Numerous online users flocked to the game platform, and brand companies started using online games to promote their products. Medical institutions have also implemented policies for healthy physical activity through online games. In order to attract more global game users, research on immersion in games is also very important to prevent users from leaving the platform. Therefore, it is necessary to analyze the common immersion elements of popular games by country and study them from the perspective of changing devices. In this paper, representative popular games of Korea, China, and Japan, which are recently called the dark horse of the game industry, are classified by devices of mobile, online PCs, and VR games and analyzed using immersion analysis tools proposed by researcher. Next, the representative metaverse games of the Web 3.0 era were analyzed, and their characteristics were compared with the results of the immersion analysis between Korea, China, and Japan, and proposed as game immersion elements suitable for metabus games.

With respect to the advent of realistic media, the so-called future cinema theory or the theory of the end of cinema that cinema will be transformed by the advent of realistic media, or that realistic media will expel film media and replace them are being put forward as if they were orthodoxy. Therefore, this paper argues that the future cinematic theory and the theory of the end of cinema the cinema are inferred from the false premise that the superiority and inferiority relationship and substitute relationship established only between video media of the same type with the same properties are also established even between realistic media and film media, which have different properties. And it is concluded that realistic media and film media will coexist because they have different properties, functions, uses, and utilities.

Research in the field of computer vision based on deep learning is being actively conducted. However, deep learning-based models have vulnerabilities in adversarial attacks that increase the model's misclassification rate by applying adversarial perturbation. In particular, in the case of FGSM, it is recognized as one of the effective attack methods because it is simple, fast and has a considerable attack success rate. Meanwhile, as one of the efforts to visualize deep learning models, Grad-CAM enables visual explanation of convolutional neural networks. In this paper, I propose a method to generate adversarial examples with high attack success rate by applying Grad-CAM to FGSM. The method chooses fixels, which are closely related to labels, by using Grad-CAM and add perturbations to the fixels intensively. The proposed method has a higher success rate than the FGSM model in the same perturbation for both targeted and untargeted examples. In addition, unlike FGSM, it has the advantage that the distribution of noise is not uniform, and when the success rate is increased by repeatedly applying noise, the attack is successful with fewer iterations.

Outlier detection refers to the task of detecting data that deviate significantly from the normal data distribution. Most outlier detection methods compute an outlier score which indicates the degree to which a data sample deviates from normal. However, setting a threshold for an outlier score to determine if a data sample is outlier or normal is not trivial. In this paper, we propose a binary prediction method for outlier detection based on spectral clustering and one-class SVM ensemble. Given training data consisting of normal data samples, a clustering method is performed to find clusters in the training data, and the ensemble of one-class SVM models trained on each cluster finds the boundaries of the normal data. We show how to obtain a threshold for transforming outlier scores computed from the ensemble of one-class SVM models into binary predictive values. Experimental results with high dimensional text data show that the proposed method can be effectively applied to high dimensional data, especially when the normal training data consists of different shapes and densities of clusters.