A Study on Establishing Monitoring Standards for Coastal Debris using Remote Sensing and Artificial Intelligence Solutions

As problems regarding the impact of marine debris on the marine environment emerge, research is being conducted to address the management of various types of marine debris, including coastal, floating, and sedimentation. In the case of Korea, which has a peninsula-shaped topography surrounded by the sea on three sides, significant damage from coastal debris occurs every year, and a national-level management plan is being prepared for this. Recently, to investigate the various types of beaches in Korea, coastal debris surveys using remote sensing devices such as UAV, CCTV, satellites, and mobile devices are being conducted. However, unlike human surveys where clear standards for coastal debris surveys using remote sensing devices are defined, research is needed to establish uniform standards. Therefore, in this study, we examined the optimal coastal debris survey method for each device and the application of artificial intelligence for automated recognition to establish the consistency of coastal debris survey methods using remote sensing devices and the validity of survey standards. The research method involved analyzing guidelines and previous research cases and applying visual intelligence using data collected from actual sea areas. First, we decided to use UAV, CCTV, and mobile devices for coastal debris investigation and research using existing marine debris monitoring guidelines, and derived data collection methods for each device by referencing the human-collected coastal debris investigation method. Additionally, for the analysis of previous research cases, a meta-analysis was performed using the above research papers on coastal debris using remote sensing devices, and trends in the field of coastal debris investigation by device were confirmed. In addition, coastal debris data were collected using remote sensing devices in various actual sea areas (stones, plants, sand, etc.) to qualitatively confirm the degree of object recognition and confirm differences by geographical characteristics of the coastal. Lastly, artificial intelligence provided visual information. We conducted a review of automated recognition methods other than human-eye recognition by applying it to the field of visual intelligence that uses information delivered using. Finally, we verified the unique characteristics of each remote sensing device, such as spatial resolution and available time, and extracted information such as the size that can recognize objects and the degree of color recognition. We also extracted coastal debris suitable for the device, such as the number of survey personnel required, monitoring cycle, and suitable target waters. Monitoring considerations were derived. By combining coastal debris survey considerations, we proposed criteria for coastal debris surveys, including device selection according to survey purpose and data collection methods for each device according to survey method, and used the proposed standards to collect coastal debris survey data using visual intelligence. As a result of its application, it had a positive impact. The results of this study are meaningful in suggesting guidelines for selecting a survey method according to Korean coastal areas with diverse geographical characteristics and debris distribution, and are expected to be helpful in supporting various decision-making for coastal surveys.