Automated Marine Biomass Recognition in Uncontrolled Underwater Environment for Marine Biomass Evaluations in Decommissioning Options Assessment

Mohd Izzat Mohd Thiyahuddin, Ghazali Ithnin, M Ismail Samsuddin,M Zhafran M Sulaiman,M Ikhranizam M Ros, M Redzuan Abdul Rahman

Day 2 Wed, December 04, 2019(2019)

引用 1|浏览0
暂无评分
摘要
Abstract Each topside structure due to be decommissioned requires a thorough case-by-case evaluation for best abandonment options. With the prospect of conducting decommissioning campaigns to reduce cost, decommissioning engineers still need to conduct multiple individual decommissioning options assessments (DOA) ahead of the bundled campaigns. Hence, it is necessary to equip decommissioning engineers with new technologies to expedite comparative assessments and evaluations. With more decommissioning being completed in campaigns, engineers require a fast automated technique to quantify and categorize marine biomass surrounding offshore structures for DOA evaluations. This tool aims to provide accurate count for marine biomass to support the decision-making processes of fisheries, marine conservation proponents and scientists in decommissioning option assessments. Researcher from PETRONAS utilizes of available ROV inspection videos from PETRONAS assets and employ image processing to detect and track marine biomass along the jackets. In this paper, a method of biomass tracking using automated scanning of video data is presented. The approach describes the initial research carried out to detect fish in the deep-sea environment video. The sequence used in this project consist of three components which is preprocess, detect and track marine lives. The tool employed produces encouraging results of detecting medium-large marine biomass from the video. Smaller minnows proved to be a challenge due to the quality of video and existence of marine snow. Once detected, the algorithm creates bounding boxes to monitor the marine surrounding the jacket leg. Results from the study showed that the precision rated at between 63% to 75% compared to manual tracking. Further improvement to this tool will include image processing alongside deep learning to increase its accuracy. The outcome of this tool will be useful during regulatory compliance & permitting phase as well as during the post remediation phase in the decommissioning value chain.
更多
查看译文
关键词
marine biomass evaluations,marine biomass,uncontrolled underwater environment
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要