| DE LEON AGUILAR SERGIO | D, 中間発表 | ユビキタスコンピューティングシステム | 安本 慶一 | 向川 康博 | 諏訪 博彦 | 松田 裕貴 | |
|
title: Mobile AR Interfaces for Instruction-based Guidelines for the General Public
abstract: Guidelines are a widely spread media used to deliver expertise knowledge and the techniques to apply it, to people with less expertise in said area. Regardless, guidelines such as government advice for preparedness are not usually read by their intended public. Augmented Reality (AR) interfaces are known to improve knowledge transfer in education, the industry, and assistance to the elderly. However, they are authored with a specific public in mind, not the general public. In our line of research, we have created a novel AR-assisted disaster prevention guideline that utilizes object detection models to guide our users to the targets of disaster preparedness advice. Finding a lower performance by minors and seniors compared to adults, we found seniors having difficulties exploring the experimental area and finding the exposed advice. To improve their experience without obstructing the experience of more tech-savvy users, by analyzing physiological signals to detect concentration and stress, we propose a novel method to identify the right timing to provide contextualized help interventions to the user. This study highlights the importance of holistic guidance alternatives for the less technology-aware population, contributing to the under-explored area of AR interfaces for the general public. language of the presentation: English | |||||||
| MANYESELA YONA ZAKARIA | D, 中間発表 | ユビキタスコンピューティングシステム | 安本 慶一 | 向川 康博 | 石山 塁 | 石寺 永記 | 諏訪 博彦 |
|
Title: Fake
Drug Package Detection Through Image Processing Abstract:
The widespread of counterfeit drugs in developing countries such as Tanzania
significantly impacts public health and leads to numerous preventable deaths of
consumers, as they often lack practical tools to authenticate drug products.
This study presents a system that assists in detecting discrepancies in drug
packaging using image recognition and web intelligence. The system compares
user-captured images with a local database of drug packages and, if necessary,
utilizes online resources through web crawling to continuously improve the
database. Advanced image feature matching algorithms are then used to identify
and visually highlight the design variations between user images and reference
images from both local and web-crawled sources. These visualized differences
help consumers easily spot potential counterfeits, providing an effective
method for pharmaceutical packaging analysis and contributing to public health
safety efforts. Language of presentation: English | |||||||
| 片岡 優衣奈 | M, 2回目発表 | 光メディアインタフェース | 向川 康博 | 安本 慶一 | 舩冨 卓哉 | 藤村 友貴 | 北野 和哉 |
|
title:Spatio-temporal advection modeling on a sphere for real-time precipitation forecasting
abstract: In real-time precipitation forecasting, precipitation observations by radiometers do not cover the entire globe everytime, and rainfall conditions in unobserved areas are supplemented by advection estimated from infrared images of clouds. However, the conventional cross-correlation-based advection estimation method does not cope with rotation and large changes, and can be highly erroneous. To improve the accuracy of advection estimation, a reliable correspondence point-based method is used to estimate the advection field whole globe. Since equidistant cylindrical projection commonly used for infrared images of clouds distorts the shape of the region nonlinearly near the poles and makes it difficult to obtain the corresponding points, I estimate the corresponding points on the tangent plane that preserves isotropy. In order to model a continuous advection field over the entire globe, a spatio-temporal model was constructed by considering advection as a rotation on a sphere, and experiments were conducted to evaluate its accuracy. language of the presentation:Japanese 発表題目:リアルタイム降水予測のための球面上での時空間移流モデルの構築 発表概要: リアルタイム降水予測では、放射計による降水観測は地球全域を毎時カバーしておらず、未観測領域の降雨状況は雲の赤外画像から推定した移流を用いて補完されている。しかし、従来の相互相関ベースの移流推定手法では回転や大きな変化に対応しておらず、大きく誤ることがある。移流の推定精度の向上を目的として、信頼性の高い対応点ベース手法で全球での移流場を推定する。雲の赤外画像でよく用いられる正距円筒図法では極付近で領域の形が非線形に歪み，対応点が得られにくいため、等方性を保つ接平面上で対応点の推定を行う．全球で連続的な移流の場をモデル化するため，移流を球面上での回転とみなして時空間モデルを構築し、精度評価の実験を行なった。 | |||||||