In this paper, we propose a data augmentation technique based on Convolutional Neural Networks (CNN or ConvNet) training to efficiently obtain a dataset of images containing concrete cracks. Concrete cracks usually do not have a standardized shape and have complex patterns, making it difficult to obtain images of them, and there is a risk of exposure to dangerous situations when securing data. Therefore, in this paper, we efficiently address the difficulty of dataset collection by using a data augmentation technique based on learning the direction and thickness of cracks, which is cost-effective and time-efficient. Moreover, to improve efficiency, we introduce a method of adaptively handling crack data by constructing a quadtree based on the presence of cracks. To confirm the extent of the improvement in accuracy, we conducted experiments applying the crack detection algorithm to various scenes, and the accuracy was improved in all scenes when measured by IoU (Intersection over union) accuracy. When the algorithm was performed without augmenting the crack data, the false detection rate was about 25%. However, when we augmented the data using our method, the false detection rate significantly decreased to 3%.

In this paper, we propose a new interface to control VR(Virtual reality) contents, games, and animations in real-time using the user’s breath and the acceleration sensor of a mobile device. Although interaction techniques are very important in VR and physically-based animations, UI(User interface) methods using different types of devices or controllers have not been covered. Most of the proposed interaction techniques have focused on screen touch and motion recognition. The direction of the breath is calculated using the position and angle between the user and the mobile device, and the control position to handle the contents is determined using the acceleration sensor built into the mobile device. Finally, to remove the noise contained in the input breath, the magnitude of the wind is filtered using a kernel modeling a pattern similar to the actual breath. To demonstrate the superiority of this study, we produced real-time interaction results by applying the breath as an external force of VR contents, games, and animations.

This paper presents a particle‐grid method to simulate the generation of opaque ice which has air bubbles in it. Water temperature is diffused over a grid, and the exchange of dissolved air between ice and water particles is simulated. We render a particle as an air bubble if it has sufficient air. Otherwise, it is treated as a cloudy volume by distributing air into dissolved air field when the final state has been reached. In addition, our method includes a model in which heat transfer rate may change across the grid. Unlike previous models which could generate an ice volume of only fixed shapes, our approach uses signed distance function (SDF) to generate opaque ice volumes stored in containers of various geometric shapes and can render needle‐shaped or egg‐shaped bubbles.