• IEMEK Journal of Embedded Systems and Applications
• /
• v.16 no.5
• /
• pp.179-186
• /
• 2021

For stereo matching based on deep learning, the design of network structure is crucial to the calculation of matching cost, and the time-consuming problem of convolutional neural network in image processing also needs to be solved urgently. In this paper, a method of stereo matching using sparse loss volume in parallax dimension is proposed. A sparse 3D loss volume is constructed by using a wide step length translation of the right view feature map, which reduces the video memory and computing resources required by the 3D convolution module by several times. In order to improve the accuracy of the algorithm, the nonlinear up-sampling of the matching loss in the parallax dimension is carried out by using the method of multi-category output, and the training model is combined with two kinds of loss functions. Compared with the benchmark algorithm, the proposed algorithm not only improves the accuracy but also shortens the running time by about 30%.

• Journal of Information Processing Systems
• /
• v.17 no.2
• /
• pp.227-241
• /
• 2021

Object detection techniques based on deep learning such as YOLO have high detection performance and precision in a single channel video stream. In order to expand to multiple channel object detection in real-time, however, high-performance hardware is required. In this paper, we propose a novel back-end server framework, a real-time AI vision platform (RAVIP), which can extend the object detection function from single channel to simultaneous multi-channels, which can work well even in low-end server hardware. RAVIP assembles appropriate component modules from the RODEM (real-time object detection module) Base to create per-channel instances for each channel, enabling efficient parallelization of object detection instances on limited hardware resources through continuous monitoring with respect to resource utilization. Through practical experiments, RAVIP shows that it is possible to optimize CPU, GPU, and memory utilization while performing object detection service in a multi-channel situation. In addition, it has been proven that RAVIP can provide object detection services with 25 FPS for all 16 channels at the same time.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2013.05a
• /
• pp.1107-1110
• /
• 2013

컴퓨터와 멀티미디어, 네트워크 기술의 발달을 바탕으로 웹 기반 교육이 출현하였고 기존 교실 중심수업의 한계를 뛰어 넘는 교육적 잠재성에 교육 운영자 및 학습자는 많은 기대를 하게 되었다. 웹을 이용한 학습은 자기주도적인 학습이 가능하고 학습자와 교수자 간의 상호 보완적인 학습이 이루어질 수 있다는 장점이 있기 때문이다. 이러한 웹 기반 학습을 효과적으로 운영하기 위해서는 개별 교육기관 자체적으로 개발한 학습관리시스템(Learning Management System;LMS)이 필요하다. 학습자들에게 양질의 온라인 콘텐츠를 제공하고자 하는 교사들을 위해 개발된 학습관리시스템 중 하나가 Moodle LMS이다. Moodle LMS는 소스가 공개되어 있어 누구나 필요한 사람은 소스를 개선, 확장, 활용할 수 있다. 따라서, 본 논문은 Moodle LMS를 활용하여 학습자가 원활하게 토론에 참여할 수 있는 토론 시스템을 설계하였다.

• The Journal of Korea Robotics Society
• /
• v.18 no.3
• /
• pp.260-270
• /
• 2023

We introduce a novel Visual Inertial Odometry (VIO) algorithm designed to improve the performance of thermal-inertial odometry. Thermal infrared image, though advantageous for feature extraction in low-light conditions, typically suffers from a high noise level and significant information loss during the 8-bit conversion. Our algorithm overcomes these limitations by approximating a 14-bit raw pixel histogram into a Gaussian mixture model. The conversion method effectively emphasizes image regions where texture for visual tracking is abundant while reduces unnecessary background information. We incorporate the robust learning-based feature extraction and matching methods, SuperPoint and SuperGlue, and zero velocity detection module to further reduce the uncertainty of visual odometry. Tested across various datasets, the proposed algorithm shows improved performance compared to other state-of-the-art VIO algorithms, paving the way for robust thermal-inertial odometry.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.11a
• /
• pp.1-3
• /
• 2022

최근 기존의 영상 압축 파이프라인 대신 신경망의 종단 간 학습을 통해 압축을 수행하는 알고리즘의 연구가 활발히 진행되고 있다. 본 논문은 종단 간 학습 기반 공간적 스케일러블 압축 기술을 제안한다. 보다 구체적으로 본 논문은 신경망의 각 계층에서 하위 계층의 학습된 특징 (feature)을 융합하여 상위 계층으로 전달하는 다중 스케일 특징 융합 (multi-scale feature fusion) 모듈을 도입해 상위 계층이 더욱 풍부한 특징 정보를 학습하고 계층 사이의 특징 중복성을 더욱 잘 제거할 수 있도록 한다. 기존 방법 대비 향상 계층(enhancement layer)에서 1.37%의 BD-rate가 향상된 결과를 볼 수 있다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.07a
• /
• pp.453-454
• /
• 2021

본 논문에서는 딥아이(DIY) 블록 프로그래밍과 라즈베리파이의 피지컬 컴퓨팅을 활용해 엑츄에이터와 센서를 제어하고 센서를 통해 수집한 데이터를 전처리해 인공지능에 활용함으로써 효율적인 인공지능 교육 방식을 제안한다. 해당 방식은 블록코딩 방식을 사용함으로써 문자코딩 대비 오타을 줄이고 문법 구애율을 낮춤으로써 프로그래밍 입문자의 구문적 어려움을 최소화하고 개념과 전략적 학습을 극대화한다. 블록프로그래밍 사용언어로 파이썬을 채택해 입문자의 편의를 도모하고 파일처리, 크롤링, csv데이터 추출을 통해 인공지능 교육에 활용한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.01a
• /
• pp.193-196
• /
• 2021

정보산업의 발달로 동영상 콘텐츠들이 폭발적으로 늘고 있다. 동영상 시청과 동시에 학습을 위해 필기를 할 경우에 화면 분할로 인한 시선 분산으로 강의에 집중하기에 한계가 있다. 이러한 한계를 극복하기 위해 투명도 조정, 최상위 고정, 과목과 주차별 분류 등을 활용하여 학습효과를 극대화함으로써 시청중인 동영상 콘텐츠와 같은 속도로 같은 곳을 바라보며 필기가 가능한 투명한 메모 모듈을 구현하였다.

• Educational Technology International
• /
• v.14 no.1
• /
• pp.1-26
• /
• 2013

This study is to identify whether three types of online instruction (text-based, video-based, and video-based instruction without text) and age category have a different influence on students' comprehension and motivation. Online students were randomly assigned to one of six groups, and they attended two-week online lectures via Course Management System. The comprehension test and the short form of IMMS were implemented when 114 participants accomplished two-week online lectures. The results revealed that using instructional video in online instruction is more effective instructional medium than text only in order to promote learner's motivation. Besides, older adults aged 41-60 are significantly different from younger adults (21-40 years old) in terms of students' comprehension. Furthermore, three types of online instructions are likely to be influenced by age category.

• International Journal of Advanced Culture Technology
• /
• v.11 no.4
• /
• pp.1-8
• /
• 2023

Objective: Despite the recommendation of the use of high-fidelity simulation (HFS) in interprofessional education (IPE), there is little known about its work for nursing students. Thus, this study aimed to explore nurse educators' perceptions and experiences in developing and implementing the HFS-based IPE for nursing and medical students. Methods: This study used a case study, using reflective filed notes. Results: Nursing educators perceived HFS as an effective educational approach to engaging nursing and medical students actively in interprofessional collaborative practice (ICP) experiences and in evaluating their actual performance on it. In terms of their perspectives on the elements necessary for effective HFS-based IPE, four themes were identified: collaborative learning, co-facilitating debriefing, leadership commitment and active faculty involvement.

• The Journal of Korea Robotics Society
• /
• v.19 no.2
• /
• pp.196-202
• /
• 2024

The ability to understand given environments and plan a sequence of actions leading to goal state is crucial for personal service robots. With recent advancements in deep learning, numerous studies have proposed methods for state representation in planning. However, previous works lack explicit information about relationships between objects when the state observation is converted to a single visual embedding containing all state information. In this paper, we introduce graph-based state representation that incorporates both object and relationship features. To leverage these advantages in addressing the task planning problem, we propose a Graph Neural Network (GNN)-based subgoal prediction model. This model can extract rich information about object and their interconnected relationships from given state graph. Moreover, a search-based algorithm is integrated with pre-trained subgoal prediction model and state transition module to explore diverse states and find proper sequence of subgoals. The proposed method is trained with synthetic task dataset collected in simulation environment, demonstrating a higher success rate with fewer additional searches compared to baseline methods.