• 문화기술의 융합
• /
• 제7권4호
• /
• pp.641-652
• /
• 2021

자율 주행 중에 많은 운전자는 안전하다고 판단되는 상황에서 운전 외 다른 활동을 수행할 것으로 예상한다. 본 연구는 반 자율주행 차량의 안전한 상황에 대한 알림의 제어권 전환 시 상황 인식 수준과 운전 수행에 미치는 영향과 주관적 평가를 살펴보았다. 실험 1에서 통제조건(경고음), 주의 알림, 안전 알림, 모든 알림 조건에 대하여 자율 주행 차량의 영상을 보고 상황 인식 수준과 주관적 평가를 진행하였다. 그 결과 안전한 상황 알림 조건에서 상황 인식 수준이 가장 높았으며 만족도와 즐거움 척도에서 높은 평가를 받았고, 불신과 불쾌함에서는 낮은 평가를 보였다. 실험 2에서는 자율주행 차량 시뮬레이터를 이용하여 실제 운전 수행, 상황 인식 수준과 주관적 평가를 진행하였다. 그 결과 운전 수행에서 안전한 알림 조건에서 가장 높은 수행을 보였으며, 더 위험이 낮다고 주관적으로 평가하였다. 본 연구는 반자율주행 차량에서 안전한 상황에 대한 알림이 운전자의 만족도와 운전 수행을 개선할 수 있음을 보여줘 불쾌한 경험을 줄이면서 안전한 자율 주행 시스템을 디자인하는 데 도움이 될 수 있을 것으로 보인다.

• Smart Structures and Systems
• /
• 제31권4호
• /
• pp.335-349
• /
• 2023

Bridges constantly undergo deterioration and damage, the most common ones being concrete damage and exposed rebar. Periodic inspection of bridges to identify damages can aid in their quick remediation. Likewise, identifying components can provide context for damage assessment and help gauge a bridge's state of interaction with its surroundings. Current inspection techniques rely on manual site visits, which can be time-consuming and costly. More recently, robotic inspection assisted by autonomous data analytics based on Computer Vision (CV) and Artificial Intelligence (AI) has been viewed as a suitable alternative to manual inspection because of its efficiency and accuracy. To aid research in this avenue, this study performs a comparative assessment of different architectures, loss functions, and ensembling strategies for the autonomous segmentation of bridge components and damages. The experiments lead to several interesting discoveries. Nested Reg-UNet architecture is found to outperform five other state-of-the-art architectures in both damage and component segmentation tasks. The architecture is built by combining a Nested UNet style dense configuration with a pretrained RegNet encoder. In terms of the mean Intersection over Union (mIoU) metric, the Nested Reg-UNet architecture provides an improvement of 2.86% on the damage segmentation task and 1.66% on the component segmentation task compared to the state-of-the-art UNet architecture. Furthermore, it is demonstrated that incorporating the Lovasz-Softmax loss function to counter class imbalance can boost performance by 3.44% in the component segmentation task over the most employed alternative, weighted Cross Entropy (wCE). Finally, weighted softmax ensembling is found to be quite effective when used synchronously with the Nested Reg-UNet architecture by providing mIoU improvement of 0.74% in the component segmentation task and 1.14% in the damage segmentation task over a single-architecture baseline. Overall, the best mIoU of 92.50% for the component segmentation task and 84.19% for the damage segmentation task validate the feasibility of these techniques for autonomous bridge component and damage segmentation using RGB images.

• Smart Structures and Systems
• /
• 제31권4호
• /
• pp.351-363
• /
• 2023

Visual structural inspections are an inseparable part of post-earthquake damage assessments. With unmanned aerial vehicles (UAVs) establishing a new frontier in visual inspections, there are major computational challenges in processing the collected massive amounts of high-resolution visual data. We propose twin deep learning models that can provide accurate high-resolution structural components and damage segmentation masks efficiently. The traditional approach to cope with high memory computational demands is to either uniformly downsample the raw images at the price of losing fine local details or cropping smaller parts of the images leading to a loss of global contextual information. Therefore, our twin models comprising Trainable Resizing for high-resolution Segmentation Network (TRS-Net) and DmgFormer approaches the global and local semantics from different perspectives. TRS-Net is a compound, high-resolution segmentation architecture equipped with learnable downsampler and upsampler modules to minimize information loss for optimal performance and efficiency. DmgFormer utilizes a transformer backbone and a convolutional decoder head with skip connections on a grid of crops aiming for high precision learning without downsizing. An augmented inference technique is used to boost performance further and reduce the possible loss of context due to grid cropping. Comprehensive experiments have been performed on the 3D physics-based graphics models (PBGMs) synthetic environments in the QuakeCity dataset. The proposed framework is evaluated using several metrics on three segmentation tasks: component type, component damage state, and global damage (crack, rebar, spalling). The models were developed as part of the 2^nd International Competition for Structural Health Monitoring.

• 국제학술발표논문집
• /
• The 5th International Conference on Construction Engineering and Project Management
• /
• pp.279-286
• /
• 2013

This paper introduces a new method for identification of building energy performance problems. The presented method is based on automated analysis and visualization of deviations between actual and expected energy performance of the building using EPAR (Energy Performance Augmented Reality) models. For generating EPAR models, during building inspections, energy auditors collect a large number of digital and thermal imagery using a consumer-level single thermal camera that has a built-in digital lens. Based on a pipeline of image-based 3D reconstruction algorithms built on GPU and multi-core CPU architecture, 3D geometrical and thermal point cloud models of the building under inspection are automatically generated and integrated. Then, the resulting actual 3D spatio-thermal model and the expected energy performance model simulated using computational fluid dynamics (CFD) analysis are superimposed within an augmented reality environment. Based on the resulting EPAR models which jointly visualize the actual and expected energy performance of the building under inspection, two new algorithms are introduced for quick and reliable identification of potential performance problems: 1) 3D thermal mesh modeling using k-d trees and nearest neighbor searching to automate calculation of temperature deviations; and 2) automated visualization of performance deviations using a metaphor based on traffic light colors. The proposed EPAR v2.0 modeling method is validated on several interior locations of a residential building and an instructional facility. Our empirical observations show that the automated energy performance analysis using EPAR models enables performance deviations to be rapidly and accurately identified. The visualization of performance deviations in 3D enables auditors to easily identify potential building performance problems. Rather than manually analyzing thermal imagery, auditors can focus on other important tasks such as evaluating possible remedial alternatives.

• 한국작물학회:학술대회논문집
• /
• 한국작물학회 2022년도 추계학술대회
• /
• pp.80-80
• /
• 2022

Estimation of heading date of paddy rice is laborious and time consuming. Therefore, automatic estimation of heading date of paddy rice is highly essential. In this experiment, deep learning classification models were used to classify two difference categories of rice (vegetative and reproductive stage) based on the panicle initiation of paddy field. Specifically, the dataset includes 444 slanted view images belonging to two categories and was then expanded to include 1,497 images via IMGAUG data augmentation technique. We adopt two transfer learning strategies: (First, used transferring model weights already trained on ImageNet to six classification network models: VGGNet, ResNet, DenseNet, InceptionV3, Xception and MobileNet, Second, fine-tuned some layers of the network according to our dataset). After training the CNN model, we used several evaluation metrics commonly used for classification tasks, including Accuracy, Precision, Recall, and F1-score. In addition, GradCAM was used to generate visual explanations for each image patch. Experimental results showed that the InceptionV3 is the best performing model in terms of the accuracy, average recall, precision, and F1-score. The fine-tuned InceptionV3 model achieved an overall classification accuracy of 0.95 with a high F1-score of 0.95. Our CNN model also represented the change of rice heading date under different date of transplanting. This study demonstrated that image based deep learning model can reliably be used as an automatic monitoring system to detect the heading date of rice crops using CCTV camera.

• 대한원격탐사학회지
• /
• 제40권1호
• /
• pp.57-69
• /
• 2024

Crop classification plays a vitalrole in monitoring agricultural landscapes and enhancing food production. In this study, we explore the effectiveness of Long Short-Term Memory (LSTM) models for crop classification, focusing on distinguishing between apple and rice crops. The aim wasto overcome the challenges associatedwith finding phenology-based classification thresholds by utilizing LSTM to capture the entire Normalized Difference Vegetation Index (NDVI)trend. Our methodology involvestraining the LSTM model using a reference site and applying it to three separate three test sites. Firstly, we generated 25 NDVI imagesfrom the Sentinel-2A data. Aftersegmenting study areas, we calculated the mean NDVI values for each segment. For the reference area, employed a training approach utilizing the NDVI trend line. This trend line served as the basis for training our crop classification model. Following the training phase, we applied the trained model to three separate test sites. The results demonstrated a high overall accuracy of 0.92 and a kappa coefficient of 0.85 for the reference site. The overall accuracies for the test sites were also favorable, ranging from 0.88 to 0.92, indicating successful classification outcomes. We also found that certain phenological metrics can be less effective in crop classification therefore limitations of relying solely on phenological map thresholds and emphasizes the challenges in detecting phenology in real-time, particularly in the early stages of crops. Our study demonstrates the potential of LSTM models in crop classification tasks, showcasing their ability to capture temporal dependencies and analyze timeseriesremote sensing data.While limitations exist in capturing specific phenological events, the integration of alternative approaches holds promise for enhancing classification accuracy. By leveraging advanced techniques and considering the specific challenges of agricultural landscapes, we can continue to refine crop classification models and support agricultural management practices.

• 대한의용생체공학회:의공학회지
• /
• 제44권5호
• /
• pp.315-323
• /
• 2023

The objective of this study is to identify the selective application targets for reporting on details of supply of class 1 and 2 medical devices as part of the improvement of the reports on details of supply of medical device system, and to analyze its effectiveness. Therapeutic materials covered by health insurance and secondhand medical devices were chosen based on the transparency of health insurance coverage and the management of medical device distribution. As a result, approximately 85% of groups can be excluded from the reporting requirements compared to reporting all items under Class 1 and 2 medical devices. This is expected to enhance the efficiency of supply reporting tasks. Additionally, the information on supply details managed by the regulatory authority can be utilized for statistical analysis and periodic monitoring, serving as fundamental data for the development of medical device-related policies and research in the field of medical devices.

• Journal of Information Technology Applications and Management
• /
• 제31권1호
• /
• pp.177-188
• /
• 2024

In this research, a cohort of two children, aged 7-8 years, was selected to participate in a specialized three-week training program aimed at enhancing their working memory. The program consisted of three sessions, each lasting approximately 30 minutes. The primary goal was to investigate the impact and developmental trajectory of working memory in school-aged children. Working memory plays a significant role in young children's learning and daily activities. To address the needs of this demographic, products should offer both educational and enjoyable activities that engage working memory. Digital educational tools, known for their flexibility, are suitable for both older individuals and young children. By updating software or modifying content, these tools can be effectively repurposed for young learners without extensive hardware changes, making them both cost-effective and practical. For example, memory training games initially designed for older adults can be adapted for young children by altering images, music, or storylines. Furthermore, incorporating elements familiar to children, like animals, toys, or fairy tales, can increase their engagement in these activities. Historically, working memory capabilities have been assessed predominantly through traditional intelligence tests. However, recent research questions the adequacy of these behavioral measures in accurately detecting changes in working memory. To bridge this gap, the current study utilized electroencephalography (EEG) as a more sophisticated and precise tool for monitoring potential changes in working memory after the training. The research findings were revealing. Participants showed marked improvement in their performance on n-back tasks, a standard measure for evaluating working memory. This improvement post-training strongly supports the effectiveness of the training program. The results indicate that such targeted and structured training programs can significantly enhance the working memory abilities of children in this age group, providing promising implications for educational strategies and cognitive development interventions.

• Computers and Concrete
• /
• 제33권5호
• /
• pp.535-544
• /
• 2024

In the rapidly advancing landscape of computer vision (CV) technology, there is a burgeoning interest in its integration with the construction industry. Camera calibration is the process of deriving intrinsic and extrinsic parameters that affect when the coordinates of the 3D real world are projected onto the 2D plane, where the intrinsic parameters are internal factors of the camera, and extrinsic parameters are external factors such as the position and rotation of the camera. Camera pose estimation or extrinsic calibration, which estimates extrinsic parameters, is essential information for CV application at construction since it can be used for indoor navigation of construction robots and field monitoring by restoring depth information. Traditionally, camera pose estimation methods for cameras relied on target objects such as markers or patterns. However, these methods, which are marker- or pattern-based, are often time-consuming due to the requirement of installing a target object for estimation. As a solution to this challenge, this study introduces a novel framework that facilitates camera pose estimation using standardized materials found commonly in construction sites, such as concrete forms. The proposed framework obtains 3D real-world coordinates by referring to construction materials with certain specifications, extracts the 2D coordinates of the corresponding image plane through keypoint detection, and derives the camera's coordinate through the perspective-n-point (PnP) method which derives the extrinsic parameters by matching 3D and 2D coordinate pairs. This framework presents a substantial advancement as it streamlines the extrinsic calibration process, thereby potentially enhancing the efficiency of CV technology application and data collection at construction sites. This approach holds promise for expediting and optimizing various construction-related tasks by automating and simplifying the calibration procedure.

• 대한영양사협회학술지
• /
• 제30권2호
• /
• pp.140-151
• /
• 2024

This study evaluated the effectiveness of an obesity program developed to solve 'obesity', which was selected as the top priority for urgent improvement among the dietary problems of elementary school students in Busan. The program aimed to practice four health rules every day (sleep early, eat two vegetables with each meal, reduce sugary snacks, and exercise for 30 minutes every day). The participants were trained to practice the four rules online in real time every day for three weeks, and their performance in nutrition education tasks was monitored using Padlet. The anthropometric measurements showed no change in the overall average weight before and after participating in the program, but all students grew in height (z=-6.978, P<0.001), and the number of obese students decreased significantly (z=-3.317, P<0.001). This obesity program was effective in improving height growth and obesity in elementary school students. In terms of dietary changes, after participating in the program, the frequency of vegetable consumption increased significantly (z=-4.849, P<0.001), the frequency of sweet snack consumption decreased significantly (z=-4.298, P<0.001), and the bedtime improved (z=-1.000). Therefore, the non-face-to-face, self-directed obesity program developed in this study is expected to reduce the workload of nutrition teachers carrying a heavy workload such as meal service and nutrition classes, and can be used as an efficient nutrition counseling program.