ScholarWorks@성균관대학교

초록

Recently, torrential rains and unpredictable bad weather around the world have led to an increasing frequency of flooding in dangerous areas and surrounding construction sites, resulting in drowning accidents every year. Such accidents can occur in various places, such as water holes, ditches, and reservoirs. Although many areas have CCTVs installed, they are not being fully utilized for water disaster prevention purposes at present. In this study, we try to prevent flooding and drowning accidents by estimating flooding areas based on computer vision. We propose a method to improve the accuracy of the analysis by removing perspective from images or images to maintain a high degree of similarity to the real world. In addition, to overcome the limitations of the existing marker-based area estimation method, we introduce a new approach using traffic lights and signs, which are standardized road facilities in urban areas. This allows information on potential drowning accident risk areas that people are not aware of in real time. This approach will greatly aid in proactive identification and management of various risk factors, as well as risk zones such as water holes, ditches, and reservoirs. This study explores the practical field applicability of this technical approach, which ultimately focuses on protecting human life from disasters.

키워드