Prostate cancer (PCa) is a fatal disease that occurs in men. In general, PCa cells are found in the prostate gland. Early diagnosis is the key to prevent the spreading of cancers to other parts of the body. In this case, deep learning-based systems can detect and distinguish histological patterns in microscopy images. The histological grades used for the analysis were benign, grade 3, grade 4, and grade 5. In this study, we attempt to use transfer learning and fine-tuning methods as well as different model architectures to develop and compare the models. We implemented MobileNet, ResNet50, and DenseNet121 models and used three different strategies of freezing layers techniques of fine-tuning, to get various pre-trained weights to improve accuracy. Finally, transfer learning using MobileNet with the half-layer frozen showed the best results among the nine models, and 90% accuracy was obtained on the test data set.

Lane detection is important technology for implementing ADAS or autonomous driving. Although edge detection has been typically used for the lane detection however, false detections occur frequently. To improve this problem, a deep learning based lane detection algorithm is proposed in this paper. This algorithm is mounted on an ARM-based embedded system to implement a LDW(lane departure warning). Since the embedded environment lacks computing power, the VGG-11, a lightweight model based on VGG-13, has been proposed. In order to evaluate the performance of the LDW, the test was conducted according to the test scenario of NHTSA.

PCB (Printed Circuit Board) inspection using a deep learning model requires a large amount of data and storage. When the amount of stored data increases, problems such as learning time and insufficient storage space occur. In this study, the existing object detection model is changed to a continual learning model to enable the recognition and classification of PCB components that are constantly increasing. By changing the structure of the object detection model to a knowledge distillation model, we propose a method that allows knowledge distillation of information on existing classified parts while simultaneously learning information on new components. In classification scenario, the transfer learning model result is 75.9%, and the continual learning model proposed in this study shows 90.7%.

In everyday life, recognizing people's emotions from their frames is essential and is a popular research domain in the area of computer vision. Visual emotion has a severe class imbalance in which most of the data are distributed in specific categories. The existing methods do not consider class imbalance and used accuracy as the performance metric, which is not suitable for evaluating the performance of the imbalanced dataset. Therefore, we proposed a method for recognizing visual emotion using balanced data augmentation to address the class imbalance. The proposed method generates a balanced dataset by adopting the random over-sampling and image transformation methods. Also, the proposed method uses the Focal loss as a loss function, which can mitigate the class imbalance by down weighting the well-classified samples. EfficientNet, which is the state-of-the-art method for image classification is used to recognize visual emotion. We compare the performance of the proposed method with that of conventional methods by using a public dataset. The experimental results show that the proposed method increases the F1 score by 40% compared with the method without data augmentation, mitigating class imbalance without loss of classification accuracy.

Although the object detection accuracy with still images has been significantly improved with the advance of deep learning techniques, the object detection problem with video data remains as a challenging problem due to the real-time requirement and accuracy drop with occlusion. In this research, we propose a method in pig detection for video monitoring environment. First, we determine a motion, from a video data obtained from a tilted-down-view camera, based on the average size of each pig at each location with the training data, and extract key frames based on the motion information. For each key frame, we then apply YOLO, which is known to have a superior trade-off between accuracy and execution speed among many deep learning-based object detectors, in order to get pig's bounding boxes. Finally, we merge the bounding boxes between consecutive key frames in order to reduce false positive and negative cases. Based on the experiment results with a video data set obtained from a pig farm, we confirmed that the pigs could be detected with an accuracy of 97% at a processing speed of 37fps.

Eye tracking research for upper limb disabilities is showing an effect in the aspect of device control. However, the reality is that it is not enough to perform web interaction with only eye tracking technology. In the Eye-Voice interface, a previous study, in order to solve the problem that the existing gaze tracking interfaces cause a malfunction of pointer execution, a gaze tracking interface supplemented with a voice command was proposed. In addition, the reduction of the malfunction rate of the pointer was confirmed through a comparison experiment with the existing interface. In this process, the difficulty of pointing due to the small size of the execution object in the web environment was identified as another important problem of malfunction. In this study, we propose an auto-magnification interface of objects so that people with upper extremities can freely click web contents by improving the problem that it was difficult to point and execute due to the high density of execution objects and their arrangements in web pages.

After arranging the general usability evaluation criteria of existing VUI researchers, this study verified how appropriate these criteria are for AI VUI specialized in navigation and the priority of their suitability. The VUI used in this study was analyzed through a survey from a total of 195 Chinese users after analyzing the navigation VUI used in China. As a result of the analysis, the usability evaluation criteria of the navigation VUI were extracted from three sub-factors of 'task accuracy', 'function satisfaction', and 'information reliability' in verifying conformance with general VUI evaluation criteria. With the recent advent of self-driving cars, safety and response speed are becoming very important, so Chinese users also ranked responsiveness as the top priority in VUI design, and the importance was also found to be high. Also, both men and women have the highest reactivity and the lowest multiplicity. VUI requires a convenient and natural interface to understand the intention between two objects through usability evaluation and verification in order to have effective interaction between humans and machines.

This thesis mainly studies the impact of short-term memory in VR tour games on users. The thesis is based on VR tour games and short-term memory, using the literature research method, the practical research method, and the investigation method. First, the author designs and makes VR tour games on the Beijing-Hangzhou Grand Canal, and then conducts a questionnaire survey and designs a control experiment. The experiment explores the differences of the short-term memory level of individuals between the normal environment and the VR tour game environment. It verifies whether the influential hypothesis proposed by the research is correct. Research conclusions show that: VR tour games have an impact on short-term memory. Compared with the normal environment, the subjects have better performance in the VR tour game mode and can maintain a high short-term memory level for a longer time. Its conclusions should promote the cultural propaganda of scenic spots and provide theoretical support for tourists' short-term memory of scenic spots culture.

When people enjoy something, they want to be repeated and focus on actions that are rewarded. The same applies to games. The purpose of this study is to ensure that the game continues to enjoy itself over and over again. Based on the artificial reward of the game, two reward methods of the game are studied. It finds Converged Difficulty Reward elements of action game through Confirmatory factor analysis in AMOS.

This study looked at the factors that shaped the identity of the Bupo-dong community through Dong-gye and Sangyeo-gye, mainly in the case of Bupori, Yeah, and Adong, and to identify that Dong-gye and Sangyeo-gye, played an important role in shaping and strengthening the community culture of its members. As a result of his inspection, Bupo-dong formed and maintained his belonging and identity as a member of Bupo-dong through the Dong-gye, the town's autonomy code. We were able to confirm that bupo-dong, unlike other villages where community culture is largely destroyed, has been able to continue its own community culture as a bupo people to this day, with a sense of belonging, solidarity and intimacy through the Sangyeo-gye of cooperation and relief. And it can be seen that the fact that the Burawon space, which has symbolic meaning to the people of Bupo-dong, exists without being watered down. It is important to discuss the community culture of Bupo-dong, which is sustainable, mainly in Bura-dong, which continues to be an important cultural space for bupo-dong people to meet and town events.

The research of combining virtual reality technology with the design elements of children's picture book education is a relatively new topic in virtual reality technology in recent years. Based on the combination of picture book design elements with virtual reality technology and the development of a children's picture book teaching game, this article analyzes the effectiveness of the application of virtual reality technology in children's teaching, and explores the usability of picture book design elements in teaching [1]. Through literature research methods, practical research methods and investigation research methods, this paper lucubrates the application of virtual reality technology in the design elements of children's picture book education so as to provide adequate theoretical and practical support for the research theme. The spatial positioning, vision, sound, and functional requirements of children's picture book games play a leading role in teaching. Practical statistics have proved that it is easier to promote children's mastery of teaching knowledge in a virtual environment. Moreover, use VR's game management function and setting function to solve the boringness of traditional education methods and the limitations of the teaching environment. The feasibility of game operation provides a virtual teaching platform system for children's education, and the teaching effect is remarkable.

This thesis provided basic data of the domestic leggings industry. For the purpose of comparative analysis, the sales status and consumer preference status of domestic leggings brands were selected, and the three companies, X*xymix, A*dar, and M*laware, which occupy the top positions by domestic leggings brand rankings Therefore, from 2018 to 2020, the market size of leggings items is identified with reference to the sales status and consumer preference status data, and the status of consumer preference is analyzed. As a result of the survey analysis, X*xymix, A*dar, and M*laware, which rank high in consumer preference, operate a separate men's leggings line from leggings targeting young women who mainly focus on yoga, pilates, and fitness exercises. As the age increased, the demand for leggings expanded to the middle-aged. Therefore, it is believed that the segmentation of leggings is made according to the diversifying types of exercise, and as the age group in demand for leggings is expanded, continuous research on market segmentation for aqua leggings necessary for aqua exercise that can exercise without straining the body.