• Title/Summary/Keyword: Attention Module

Search Result 245, Processing Time 0.03 seconds

A Study on the Safety Vest by Sports and Leisure Population Distribution -Focusing on Motorcycle Vest- (스포츠 레저 인구 확산에 따른 안전 상의에 관한 연구 -모터사이클 상의를 중심으로-)

  • Lee, Hyunyoung
    • Journal of Fashion Business
    • /
    • v.22 no.5
    • /
    • pp.125-136
    • /
    • 2018
  • This study intended to develop a motorcycle safe vest that can be prepared against accidents by mounting a smart module (with built-in sensor) on the safe vest in order to emphasize safety among functional aspects of the motorcycle clothing. The research method investigated professional books, prior research, and Internet data to examine the characteristics of motorcycle wear and the theoretical examination of smart wear, and analyzed the functional characteristics of the design by reviewing smart jacket and vest design cases for motorcycles currently on the market. As a results of study an interface device sensor, which contains a sensor with IMU(Intertial Measurement Unit) and CPU(Central Processing Unit), was inserted into a motorcycle top in order to draw attention to the safety of motorcycle riders. The IMU sensor attached to the vest detected the tilting motion of the rider to either left or right side to obtain data on left or right direction, sudden stop, and so forth and displayed left or right turn signal and sudden stop sign on the backplate (back) through the LED module. As for charging the device to operate LEDs, a generator, which is designed to convert the heat energy in the exhaust into electric energy, was used to efficiently self-produce the power required to operate LEDs of the top while riding.

Design, Fabrication and temperature measuring experiments of solar collecting system using a single module reflectors (단일 모듈 반사경을 이용한 태양열 집열 시스템의 설계, 제작 및 온도측정 시험연구)

  • Yang, Byeong-Soo;Yang, Woo;Seo, Tae-Il;Son, Chang-Woo
    • Design & Manufacturing
    • /
    • v.13 no.1
    • /
    • pp.19-24
    • /
    • 2019
  • Currently, the world is paying keen attention to the production of renewable energy along with environmental issues, and the share of renewable energy in the world is rising above that of nuclear power. Especially when Korea, which is heavily dependent on foreign countries, needs to reconsider its national competitiveness due to the recent high oil prices, the government's energy policy is to develop and use renewable energy that replaces fossil fuels. In particular, solar energy, the most actively studied and commercialized field of renewable energy, is the main research for solar energy and is commercialized and used. However, the efficiency of solar energy has already reached saturation. Studies are also focusing on increasing the reflectivity of solar energy to increase efficiency. Therefore, this paper proposes a solar collection system that can utilize solar energy rather than solar energy. The proposed solar heat collection system uses solar tracking systems to effectively collect solar energy, particularly those that can be easily produced using single-modular reflectors and have price competitiveness. In addition, temperature measurement experiments with temperature measuring sensors were conducted to ensure reliability in order to verify the results interpreted.

Visualization Based Building Anatomy Model for Construction Safety Education

  • Pham, Hai Chien;Le, Quang Tuan;Pedro, Akeem;Park, Chan Sik
    • International conference on construction engineering and project management
    • /
    • 2015.10a
    • /
    • pp.430-434
    • /
    • 2015
  • Safety education at the tertiary level prepares students to enter construction industry with adequate safety knowledge; then accidents can be prevented proactively. However, safety subject has not been paid adequate attention in universities and most institutional safety programs consider safety matters in isolation. Meanwhile, anatomical theory in the medicine field has been successfully adopted and proved potential advantageous in various scientific disciplines. With this regard, this study proposes a visualization based Building Anatomy Model (BAM) for construction safety education, which utilizes the anatomical theory in order to improve student's safety knowledge and practical skill. This BAM consists of two modules: 1) Knowledge Acquisition Module (KAM) aims to deliver safety knowledge to students through building anatomy models; 2) Practical Experience Module (PEM) where students safely perform construction activities by using the system to improve safety skill. The system trial is validated with virtual scenarios derived from real accidents cases. This study emphasizes the visualization based building anatomy model would be a powerful pedagogical method to provide effectively safety knowledge and practical skill for students, as a result, safety competence of students would be enhanced.

  • PDF

Few-shot Aerial Image Segmentation with Mask-Guided Attention (마스크-보조 어텐션 기법을 활용한 항공 영상에서의 퓨-샷 의미론적 분할)

  • Kwon, Hyeongjun;Song, Taeyong;Lee, Tae-Young;Ahn, Jongsik;Sohn, Kwanghoon
    • Journal of Korea Multimedia Society
    • /
    • v.25 no.5
    • /
    • pp.685-694
    • /
    • 2022
  • The goal of few-shot semantic segmentation is to build a network that quickly adapts to novel classes with extreme data shortage regimes. Most existing few-shot segmentation methods leverage single or multiple prototypes from extracted support features. Although there have been promising results for natural images, these methods are not directly applicable to the aerial image domain. A key factor in few-shot segmentation on aerial images is to effectively exploit information that is robust against extreme changes in background and object scales. In this paper, we propose a Mask-Guided Attention module to extract more comprehensive support features for few-shot segmentation in aerial images. Taking advantage of the support ground-truth masks, the area correlated to the foreground object is highlighted and enables the support encoder to extract comprehensive support features with contextual information. To facilitate reproducible studies of the task of few-shot semantic segmentation in aerial images, we further present the few-shot segmentation benchmark iSAID-, which is constructed from a large-scale iSAID dataset. Extensive experimental results including comparisons with the state-of-the-art methods and ablation studies demonstrate the effectiveness of the proposed method.

A group-wise attention based decoder for lightweight salient object detection on edge-devices (엣지 디바이스에서 객체 탐지를 위한 그룹별 어탠션 기반 경량 디코더 연구)

  • Thien-Thu Ngo;Md Delowar Hossain;Eui-Nam Huh
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2023.11a
    • /
    • pp.30-33
    • /
    • 2023
  • The recent scholarly focus has been directed towards the expeditious and accurate detection of salient objects, a task that poses considerable challenges for resource-limited edge devices due to the high computational demands of existing models. To mitigate this issue, some contemporary research has favored inference speed at the expense of accuracy. In an effort to reconcile the intrinsic trade-off between accuracy and computational efficiency, we present novel model for salient object detection. Our model incorporate group-wise attentive module within the decoder of the encoder-decoder framework, with the aim of minimizing computational overhead while preserving detection accuracy. Additionally, the proposed architectural design employs attention mechanisms to generate boundary information and semantic features pertinent to the salient objects. Through various experimentation across five distinct datasets, we have empirically substantiated that our proposed models achieve performance metrics comparable to those of computationally intensive state-of-the-art models, yet with a marked reduction in computational complexity.

Masked cross self-attentive encoding based speaker embedding for speaker verification (화자 검증을 위한 마스킹된 교차 자기주의 인코딩 기반 화자 임베딩)

  • Seo, Soonshin;Kim, Ji-Hwan
    • The Journal of the Acoustical Society of Korea
    • /
    • v.39 no.5
    • /
    • pp.497-504
    • /
    • 2020
  • Constructing speaker embeddings in speaker verification is an important issue. In general, a self-attention mechanism has been applied for speaker embedding encoding. Previous studies focused on training the self-attention in a high-level layer, such as the last pooling layer. In this case, the effect of low-level layers is not well represented in the speaker embedding encoding. In this study, we propose Masked Cross Self-Attentive Encoding (MCSAE) using ResNet. It focuses on training the features of both high-level and low-level layers. Based on multi-layer aggregation, the output features of each residual layer are used for the MCSAE. In the MCSAE, the interdependence of each input features is trained by cross self-attention module. A random masking regularization module is also applied to prevent overfitting problem. The MCSAE enhances the weight of frames representing the speaker information. Then, the output features are concatenated and encoded in the speaker embedding. Therefore, a more informative speaker embedding is encoded by using the MCSAE. The experimental results showed an equal error rate of 2.63 % using the VoxCeleb1 evaluation dataset. It improved performance compared with the previous self-attentive encoding and state-of-the-art methods.

Synchronization of Moving objects and VR images (매칭 모듈을 이용한 이동 객체와 VR 영상의 동기화)

  • Lee, Hyoun-Sup;You, Eun-Jae;Kim, Jin-deog
    • Journal of the Korea Institute of Information and Communication Engineering
    • /
    • v.21 no.7
    • /
    • pp.1435-1440
    • /
    • 2017
  • Recently, VR and AR are emerging as an area of interest in ICT. In virtual reality to feel like a real, many people are looking for them because of its charm. However, the motion of the VR image and the moving object are not synchronized exactly, and a problem of feeling nausea is also occurring. The problem should be solved by the application of Attraction in the VR system. In this paper, we propose a module that minimizes delay time by synchronizing movement of VR image and moving object. The proposed module calculates the moving distance using the direction and the acceleration sensor that the user views through the VR device. The module proposed in this paper will pay attention to the fact that the movement of the attraction moves along a fixed path, so that the accurate travel distance can be calculated.

The Design and development of online learning modules for the broadcasting content-based e-PBL (방송콘텐츠 기반 e-PBL을 위한 온라인 학습모듈 설계 및 개발)

  • Jung, Joon-Hwan
    • Journal of the Korea Society of Computer and Information
    • /
    • v.17 no.1
    • /
    • pp.105-115
    • /
    • 2012
  • This study paid attention to the possibility of broadcasting contents as a learning resource, which promoted the knowledge composition and the process of creation from the software having enriched contents not a delivering tool for the information in one direction. and tried to implement it thru e-PBL, which is one of teaching-learning models in e-learning. In order to implement the possibility of broadcasting contents in the e-PBL enviroment as a learning resource, the study focused on the design and development of e-PBL based on the broadcasting contents and found out the educational possibility for the broadcasting contents as the learning resource in e-PBL enviroment. This study focused on the design of learning module and development process which integrated strategic plans to maximize the utilization of broadcasting contents including development of online learning community. Also, This is verified by applying the learning module was to prove a differentiation.

Performance evaluation of forward osmosis (FO) hollow fiber module with various operating conditions (중공사막 모듈을 이용한 정삼투 공정에서의 운영조건 변화에 따른 성능평가)

  • Kim, Bongchul
    • Journal of Korean Society of Water and Wastewater
    • /
    • v.32 no.4
    • /
    • pp.357-361
    • /
    • 2018
  • Forward osmosis (FO) process has been attracting attention for its potential applications such as industrial wastewater treatment, wastewater reclamation and seawater desalination. Particularly, in terms of fouling reversibility and operating energy consumption, the FO process is assumed to be preferable to the reverse osmosis (RO) process. Despite these advantages, there is a difficulty in the empirical step due to the lack of separation and recovery techniques of the draw solution. Therefore, rather than using FO alone, recent developments of the FO process have adapted a hybrid system without draw solution separation/recovery systems, such as the FO-RO osmotic dilution system. In this study, we investigated the performance of the hollow fiber FO module according to various operating conditions. The change of permeate flow rate according to the flow rates of the draw and feed solutions in the process operation is a factor that increases the permeate flow rate, one of the performance factors in the positive osmosis process. Our results reveal that flow rates of draw and feed solutions affect the membrane performance, such as the water flux and the reverse solute flux. Moreover, use of hydraulic pressure on the feed side was shown to yield slightly higher flux than the case without applied pressure. Thus, optimizing the operating conditions is important in the hollow fiber FO system.

Design of High Payload Dual Arm Robot with Replaceable Forearm Module for Multiple Tasks: Human Rescue and Object Handling (임무에 따른 하박 교체형 고 가반하중 양팔로봇의 설계: 구난 및 물체 핸들링)

  • Kim, Hwisu;Park, Dongil;Choi, Taeyong;Do, Hyunmin;Kim, Doohyeong;Kyung, Jinho;Park, Chanhun
    • The Journal of Korea Robotics Society
    • /
    • v.12 no.4
    • /
    • pp.441-447
    • /
    • 2017
  • Robot arms are being increasingly used in various fields with special attention given to unmanned systems. In this research, we developed a high payload dual-arm robot, in which the forearm module is replaceable to meet the assigned task, such as object handling or lifting humans in a rescue operation. With each forearm module specialized for an assigned task (e.g. safety for rescue and redundant joints for object handling task), the robot can conduct various tasks more effectively than could be done previously. In this paper, the design of the high payload dual-arm robot with replaceable forearm function is described in detail. Two forearms are developed here. Each of forearm has quite a different goal. One of the forearms is specialized for human rescue in human familiar flat aspect and compliance parts. Other is for general heavy objects, more than 30 kg, handling with high degree of freedom more than 7.