https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/142 2026-07-04T22:06:17Z https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/212938 Title: A Semi-Supervised Framework for Road Condition Assessment: From Minor Surface Distress to Post-Disaster Failures Authors: Tsujimoto, Eda; Eom, Sunyong; Suzuki, Tsutomu Abstract: Accurate road-condition segmentation is crucial for intelligent transportation systems (ITS), enabling autonomous driving and post-disaster recovery. Yet, most prior studies focus on binary or low-class segmentation, limiting applicability to complex scenes where subtle surface degradations and large-scale obstructions coexist. To address this gap, we present a unified framework for pixel-level segmentation across 11 road-condition classes using a hybrid dataset integrating the Post-Disaster Road Dataset Japan (PDRDD-J), the Social Media Image Dataset for Disaster Road Damage Object Detection (SoDR), and the Road Damage Dataset 2022 (RDD2022-J). The defined classes include background, road, manhole cover, alligator crack, linear crack, pothole, vehicle, natural blockage, structural blockage, sinkhole, and collapsed road. To learn this label space under limited supervision, we propose F-UNet, a Feature-Level Data Extractor (FLDE)-guided extension of U-Net integrated into a standard teacher–student self-training protocol and trained through progressive pairwise specialization. Unlike conventional U-Net variants that rely solely on global multi-class training, F-UNet derives FLDE guidance through pairwise road–damage patch training and injects it during decoding to better handle class imbalance and thin, low-contrast defects. Extensive experiments demonstrate consistent gains over strong baselines, supported by high-confidence class-wise analysis, confusion-matrix analysis, and error and uncertainty-based failure-mode characterization under disaster edge cases. On a held-out test set, the teacher model achieves 0.8652 mIoU, representing a 15.5% improvement over standard U-Net and outperforming U-Net++, DeepLabV3+, SegNet, and FCN under the same supervision setting. In the semi-supervised configuration, the student model further improves to 0.8727 mIoU. The F-UNet source code is available at https://github.com/EdaTsujimoto/F-UNET 2026-05-01T00:00:00Z https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/217878 Title: 백화점 폐점이 주택시장과 지역 상업활동에 미치는 영향 - 수원 · 창원 사례를 대상으로 - Authors: 조에스더; 이명훈; 엄선용 Abstract: 본 연구는 국내 백화점 산업이 구조적 재편 국면에 진입한 가운데, 도심형 앵커시설인 백화점의 폐점이 인근 아파트 가격과 지역 상권의 지속가능성에 미치는 영향을 실증적으로 분석하였다. 아파트 실거래가와 음식점 폐업 데이터를 활용하여 주택시장에는 이중차분법(difference-in-differences, DID)과 삼중차분법(difference-in-differences-in-differences, DDD)을 적용하고, 상권 변화에는 BSTS(Bayesian structural time series) 기반 Causal Impact 모형을 활용하였다. 분석 결과, 폐점 이후 인근 아파트 가격은 전반적으로 하락하는 경향을 보였으며, 수원 사례에서는 핵심 영향권에서 추가적인 가격 하락이 나타나 공간적 이질성이 확인된 반면 창원 사례에서는 가격 하락 경향이 관찰되었으나 통계적으로 유의한 수준은 아니었다. 또한, 상권 분석에서는 음식점 폐업 증가가 나타나 지역 상업활력의 위축 가능성이 확인되었으며, 이러한 영향의 크기와 지속성은 도시의 구조적 특성과 시장 여건에 따라 상이하게 나타났다. 이는 백화점 폐점의 영향이 단일한 인과관계로 설명되기보다 지역적 맥락에 따라 달라질 수 있음을 시사한다.; This study examines the effects of department store closures on nearby housing prices and local commercial activity in South Korea, as the sector enters a phase of structural transformation marked by increasing polarization between high-performing flagship stores and underperforming non-core locations. Unlike discount stores, department stores function as high-order urban anchor facilities that integrate retail, cultural, and symbolic roles. This suggests that their closures may generate broader urban impacts beyond a simple reduction in retail supply. Using apartment transaction data and restaurant closure records, this study applies difference-in-differences (DID) and difference-in-difference-in-differences (DDD) models to analyze housing market impacts and employs a Bayesian structural time series-based causal impact model to estimate counterfactual changes in local retail activity. The results indicate a general downward trend in housing prices following department store closures. In Suwon, the DID estimates demonstrate no statistically significant average effect, whereas the DDD model identifies an additional price decline within the core impact zone, particularly in areas closer to the closed department stores, suggesting localized spatial effects. By contrast, in Changwon, although the estimated coefficients consistently point to a negative direction, no statistically significant housing price decline is identified. Regarding retail activity, the findings suggest an increase in restaurant closures after the shutdown, implying a contraction in local commercial vitality. However, both the magnitude and persistence of these effects vary across cases, reflecting differences in urban structures and market conditions. Overall, the results suggest that department stores play a meaningful role as urban anchor facilities, but the impacts of closures are not uniform and are shaped by broader urban development trajectories and local contexts. This study provides empirical evidence to inform urban policies and post-closure redevelopment strategies that account for regional heterogeneity. 2026-04-01T00:00:00Z https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/210386 Title: Designing age-friendly streetscapes: Assessing environmental risk factors for falls using street view images Authors: Hong, Andy; Ha, Jaehyun; Ki, Donghwan; Choi, Dong-ah Abstract: As cities worldwide strive to become more age-friendly in response to rapidly aging populations, outdoor falls among older adults remain a persistent challenge that undermines these initiatives. The micro-scale environmental determinants of these falls remain inadequately explored, creating a significant gap in evidence-based urban design for inclusive urban environments. This study investigates the association between built environment features and outdoor fall risks among older adults using data from 6302 emergency dispatch cases in Jeonbuk Province, South Korea. By leveraging deep learning-based computer vision and a zero-inflated Poisson model, we analyze half a million street view images to identify micro-scale streetscape features associated with outdoor fall incidents. Our findings indicate that older adults are more likely to fall in areas with high population density and mixed-use land. While these environmental characteristics show higher fall incident rates, certain streetscape features, such as curbs and brick surfaces, can mitigate fall risks, while asphalt and concrete surfaces may exacerbate them. These findings demonstrate that environments promoting walking require enhanced fall risk management through appropriate design considerations for older adults. By recognizing older adults as ‘mobility minorities’ with specific needs, targeted micro-scale streetscape interventions, such as surface materials and curb designs, can enhance safety while preserving walkability benefits, supporting more inclusive urban environments. 2026-03-01T00:00:00Z https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/213192 Title: Exploring network scale separation strategies for car-bicycle integration Authors: Liu, Liling; Eom, Sunyong; Suzuki, Tsutomu Abstract: This study investigates strategies to mitigate car–bicycle conflicts in mixed traffic and their impacts on traffic speed and safety. It proposes and evaluates an approach that separates bicycles and cars onto different roads in a network. Various scenarios were compared with a baseline, accounting for traffic volume, modal share, and road hierarchy where bicycles and cars are separated. The performance of each scenario was evaluated from the perspectives of motorists and cyclists, considering car and bicycle efficiency across different trip lengths, as well as cycling stress levels assessed using the Level of Traffic Stress (LTS) score. The methodology involved estimating travel times using a traffic simulator and generating reachable areas for bicycles and cars. The study provides insights for designing multimodal transportation systems that consider both the benefits of shared road space and the potential advantages of separating bicycles and cars onto different roads. The main results are as follows: (1) Cars and bicycles show a trade-off relationship in transport efficiency in all network scenarios; the scenarios differ in the road hierarchy levels at which car and bicycle traffic are separated onto different roads; (2) Separating bicycles from cars on middle-class and local roads can upgrade the cycling environment, including efficiency and comfort, both on roads and at intersections; (3) To reconcile conflicts between motorized speed and cyclists’ comfort, enlarging high-hierarchy roads for car-dedicated use can be effective. 2026-03-01T00:00:00Z