Breast cancer is one of the most common cancers in women worldwide. In Korea, breast cancer is most common cancer in women followed by thyroid cancer. The purpose of this study is to evaluate the possibility of using deep - run model for segmentation of breast masses and to identify the best deep-run model for breast mass segmentation. In this study, data of patients with breast masses were collected at Asan Medical Center. We used 596 images of mammography and 596 images of gold standard. In the area of interest of the medical image, it was cut into a rectangular shape with a margin of about 10% up and down, and then converted into an 8-bit image by adjusting the window width and level. Also, the size of the image was resampled to $150{\times}150$. In Deconvolution net, the average accuracy is 91.78%. In U-net, the average accuracy is 90.09%. Deconvolution net showed slightly better performance than U-net in this study, so it is expected that deconvolution net will be better for breast mass segmentation. However, because of few cases, there are a few images that are not accurately segmented. Therefore, more research is needed with various training data.

There is a problem to measure neutral bio-signals during sleep because of inconvenience of attaching lots of sensors. In this study, we measured single electrocardiogram(ECG) signal and analyzed the correlation with sleep. After R-peak detection from ECG signal, we extracted 9 features from time and frequency domain of heart rate variability(HRV). Mean of HRV, RR intervals differing more than 50ms(NN50), and divided by the total number of all RR intervals(pNN50) have significant differences in each sleep stage. Specially, the mean HRV has an average of 87.8% accuracy in classifying sleep and awake status. In the future, the measurement ECG signal minimizes inconvenience of attaching sensors during sleep. Also, it can be substituted for the standard sleep measurement method.

In this paper, we proposed eyeglass type periorbital wrinkle treatment device for reducing and improving periorbital wrinkles using near infrared LED of 850nm wavelength and galvanic current. The proposed periorbital wrinkle treatment device is equipped with a control system based on F-PCB. It consists of eight near-infrared LEDs and four indicator LEDs for treatment of right and left periorbital wrinkles. The eyeglass frame is coated with conductive material, so galvanic current can flow to the skin of periorbital wrinkle contacted to it. One male adult in the mid-40s was allowed to use the device for 10 minutes every day for 4 weeks. After 4 weeks, image analysis using optical equipment for measuring wrinkles indicated that wrinkle indexes were reduced.

Core algorithm of deep learning Convolutional Neural Network(CNN) shows better performance than other machine learning algorithms. However, if there is not sufficient data, CNN can not achieve satisfactory performance even if the classifier is excellent. In this situation, it has been proven that the use of transfer learning can have a great effect. In this paper, we apply two transition learning methods(freezing, retraining) to three CNN models(ResNet-50, Inception-V3, DenseNet-121) and compare and analyze how the classification performance of CNN changes according to the methods. As a result of statistical significance test using various evaluation indicators, ResNet-50, Inception-V3, and DenseNet-121 differed by 1.18 times, 1.09 times, and 1.17 times, respectively. Based on this, we concluded that the retraining method may be more effective than the freezing method in case of transition learning in image classification problem.

Lane detection is a key function in developing autonomous vehicle technology. In this paper, we propose a lane marker detection algorithm robust to environmental variation targeting low cost embedded computing devices. The proposed algorithm consists of two phases: initialization phase which is slow but has relatively higher accuracy; and the tracking phase which is fast and has the reliable performance in a limited condition. The initialization phase detects lane markers using a set of filters utilizing the various features of lane markers. The tracking phase uses Kalman filter to accelerate the lane marker detection processing. In a tracking phase, we measure the reliability of the detection results and switch it to initialization phase if the confidence level becomes below a threshold. By combining the initialization and tracking phases we achieved high accuracy and acceptable computing speed even under a low cost computing resources in which we cannot use the computing intensive algorithm such as deep learning approach. Experimental results show that the detection accuracy is about 95% on average and the processing speed is about 20 frames per second with Raspberry Pi 3 which is low cost device.

The purpose of this study was to train a model for the ulna and radius bone segmentation based on Convolutional Neural Networks and to verify the segmentation model. The data consisted of 840 training data, 210 tuning data, and 200 verification data. The learning model for the ulna and radius bone bwas based on U-Net (19 convolutional and 8 maximum pooling) and trained with 8 batch sizes, 0.0001 learning rate, and 200 epochs. As a result, the average sensitivity of the training data was 0.998, the specificity was 0.972, the accuracy was 0.979, and the Dice's similarity coefficient was 0.968. In the validation data, the average sensitivity was 0.961, specificity was 0.978, accuracy was 0.972, and Dice's similarity coefficient was 0.961. The performance of deep convolutional neural network based models for the segmentation was good for ulna and radius bone.

In order to improve the performance of image classifications using Convolutional Neural Networks (CNN), applying a category hierarchy to the classification can be a useful idea. However, the visual separation of object categories is very different according to the upper and lower category levels and highly uneven in image classifications. Therefore, it is doubtable whether the use of category hierarchies for classification is effective in CNN. In this paper, we have clarified whether the image classification using category hierarchies improves classification performance, and found at which level of hierarchy classification is more effective. For experiments we divided the image classification task according to the upper and lower category levels and assigned image data to each CNN model. We identified and compared the results of three classification models and analyzed them. Through the experiments, we could confirm that classification effectiveness was not improved by reduction of number of categories in a classification model. And we found that only with the re-training method in the last network layer, the performance of lower category classification was not improved although that of higher category classification was improved.

In this paper, we propose a CNN based deep learning algorithm for semantic segmentation of images. In order to improve the accuracy of semantic segmentation, we combined pixel level object classification and image level object classification. The image level object classification is used to accurately detect the characteristics of an image, and the pixel level object classification is used to indicate which object area is included in each pixel. The proposed network structure consists of three parts in total. A part for extracting the features of the image, a part for outputting the final result in the resolution size of the original image, and a part for performing the image level object classification. Loss functions exist for image level and pixel level classification, respectively. Image-level object classification uses KL-Divergence and pixel level object classification uses cross-entropy. In addition, it combines the layer of the resolution of the network extracting the features and the network of the resolution to secure the position information of the lost feature and the information of the boundary of the object due to the pooling operation.

In this paper, we proposed a river water level detection and tracking of the river or dams based on image processing system. In past, most of the water level detection system used various water sensors. Those water sensors works perfectly but have many drawbacks such as high cost and harsh weather. Water level monitoring system helps in forecasting early river disasters and maintenance of the water body area. However, the early river disaster warning system introduces many conflicting requirements. Surveillance camera based water level detection system depends on either the area of interest from the water body or on optical flow algorithm. This proposed system is focused on water scaling area of a river or dam to detect water level. After the detection of scale area from water body, the proposed algorithm will immediately focus on the digits available on that area. Using the numbers on the scale, water level of the river is predicted. This proposed system is successfully tested on different water bodies to detect the water level area and predicted the water level.

Palm image acquisition without contact has advantages in user convenience and hygienic issues, but such images generally display more image variations than those acquired employing a contact plate or pegs. Therefore, it is necessary to develop a palmprint identification method which is robust to affine variations. This study proposes a deep learning approach which can effectively identify contactless palmprints. In general, it is very difficult to collect enough volume of palmprint images for training a deep convolutional neural network(DCNN). So we adopted an approach to use a pretrained DCNN. We designed two new DCNNs based on the VGGNet. One combines the VGGNet with SVM. The other add a shallow network on the middle-level of the VGGNet. The experimental results with two public palmprint databases show that the proposed method performs well not only contact-based palmprints but also contactless palmprints.

Vehicular Ad-Hoc Networks (VANETs) have become one of the active areas of research, standardization, and development because they have tremendous potentials to improve vehicle and road safety, traffic efficiency, and convenience as well as comfort to both drivers and passengers. However, message trustfulness is a challenge because the propagation of false message by malicious vehicles induces unreliable and ineffectiveness of VANETs, Therefore, we need a reliable reputation method to ensure message trustfulness. In this paper, we consider a vulnerability against the Sybil attack of the previous reputation systems based on blockchain and suggest a new reputation system which resists against Sybil attack on the previous system. We propose an initial authentication process as a countermeasure against a Sybil attack and provide a reliable reputation with a cooperative message delivery to cope with message omission. In addition, we use Homomorphic Commitment to protect the privacy breaches in VANETs environment.

Robots which detect human and recognize action are important factors for human interaction, and many researches have been conducted. Recently, deep learning technology has developed and learning based robot's technology is a major research area. These studies require a database to learn and evaluate for intelligent human perception. In this paper, we propose a multi-modal sensor-based image database condition considering the security task by analyzing the image database to detect the person in the outdoor environment and to recognize the behavior during the running of the robot.

For the problem that the face image of surveillance video cannot be accurately identified due to the low resolution, this paper proposes a low resolution face recognition solution based on convolutional neural network model. Convolutional Neural Networks (CNN) model for multi-scale input The CNN model for multi-scale input is an improvement over the existing "two-step method" in which low-resolution images are up-sampled using a simple bi-cubic interpolation method. Then, the up sampled image and the high-resolution image are mixed as a model training sample. The CNN model learns the common feature space of the high- and low-resolution images, and then measures the feature similarity through the cosine distance. Finally, the recognition result is given. The experiments on the CMU PIE and Extended Yale B datasets show that the accuracy of the model is better than other comparison methods. Compared with the CMDA_BGE algorithm with the highest recognition rate, the accuracy rate is 2.5%~9.9%.

The extraction and recognition of human motion characteristics need to combine biometrics to determine and judge human behavior in the movement and distinguish individual identities. The so-called biometric technology, the specific operation is the use of the body's inherent biological characteristics of individual identity authentication, the most noteworthy feature is the invariance and uniqueness. In the past, the behavior recognition technology based on the single characteristic was too restrictive, in this paper, we proposed a mixed feature which combined global silhouette feature and local optical flow feature, and this combined representation was used for human action recognition. And we will use the KTH database to train and test the recognition system. Experiments have been very desirable results.

Musculoskeletal disorders show symptoms of pain and dysfunction caused by accumulation of minute damage on nerves and blood vessels which are related to specific body parts. This is more likely to occur in the elderly. In Korea, 40.3% of the agricultural population is over 65 years old. In these cases, the portion of musculoskeletal disorders is more than 80% in acute/chronic diseases related to agricultural work. Therefore, measures for physical health care and safety prevention of elderly farmers are needed. In this study, we intend to propose a method to measure musculoskeletal health condition through the motion recognition device and VR contents. For this, we analyze the agricultural work causing musculoskeletal disorders of farmers, and suggest a motion analysis algorithm that can measure the health condition of the musculoskeletal system. In addition, we are going to propose a training method customized for farmers on the basis of VR contents in order to prevent musculoskeletal disorders.

We propose an object recognition system and experiential game contents using Kinect to maximize object recognition rate by utilizing underwater robots. we implement an ice hockey game based on object-aware interactive contents to validate the excellence of the proposed system. The object recognition system, which is a preprocessor module, is composed based on Kinect and OpenCV. Network sockets are utilized for object recognition communications between C/S. The problem of existing research, degradation of object recognition at long distance, is solved by combining the system development method suggested in the study. As a result of the performance evaluation, the underwater robot object recognized all target objects (90.49%) with 80% of accuracy from a 2m distance, revealing 42.46% of F-Measure. From a 2.5m distance, it recognized 82.87% of the target objects with 60.5% of accuracy, showing 34.96% of F-Measure. Finally, it recognized 98.50% of target objects with 59.4% of accuracy from a 3m distance, showing 37.04% of F-measure.

The purpose of this study is to design and apply a simple game that can visually experience basic concepts of game mathematics in order to teach game mathematics effectively. To do this, simple games linked with game mathematical theory are to be developed by utilizing the functions provided by Unity so that students could actively learn game mathematics. To demonstrate the plausibility of this approach, "Bouncing Ball Game" was developed to understand the concept of periodicity of trigonometric functions. As a result, students were able to effectively learn how mathematical concepts related to ball movements applied to the game.

This study examines the mechanism of how the dispositif, a dispositional concept of television, which is changing in form, Q-Sheet, is expressed as 'ideology' embedded in television. The dispositif implemented in television aims at the realization of mechanical desire through the internal disposition and makes the reality that it dazzles the public by adjusting the density of the gap and depreciates the existence of the aura itself. Instead of gaining new experience through the disposition, the public accepts the manipulated experience and falls into the illusion that it is true. In the television literacy program, the Q-sheet acts in the intervals and gaps that exist in the movement between the stacks, spreading the ideology, and the act of Channel Surfing by using the remote control is also a dispositf form of television viewing. In modern capitalist societies, television disposition are regenerating through proliferation and expansion, showing intent to dominate even art and culture through disposition, and autonomously injecting ideologies by television disposition.

This study aims to provide implications for mobile foreign language vocabulary learning app by analyzing the effects of mobile vocabulary learning app based on character animation. For this purpose, we applied the learning application designed with character animation and text, and the application designed with text only to two groups of learners, and analyzed the effect. As a result, we found that application designed with character animation and text was useful in recognition frequency and duration concerning learning. Regarding learning outcomes, we found that it is useful not only in memory but also in learning interest and motivation. This study provides implications for learning method and design development of mobile-based foreign language vocabulary learning application which actively using recently.