Ocean carbon emission prediction and management measures based on artificial intelligence remote sensing estimation in the context of carbon neutrality.

With rapid economic development, the gradual deterioration of the natural environment has posed unprecedented challenges to human social civilization. The marine economy, as an important part of economic development, is the breakthrough of economic transformation for many coastal countries. Additionally, green development and environmental impact assessment have become the focus of research in these countries. This study employs remote sensing technology, an efficient observational method, to significantly enhance the efficiency of ocean information observation. It investigates ocean carbon emissions within the framework of carbon neutrality. First, we identified the ships along the coastline based on marine remote sensing information through the YOLO (you only look once) framework. Second, we applied the LSTM (long short-term memory) method to combine the target identification results and the historical data of carbon emissions to complete the corresponding carbon emission data fitting. Finally, carbon emission data from the past three years in the offshore area of Dalian were used to make accurate predictions. The results suggested that the recognition rate of the proposed target detection method could reach 88%, and the LSTM method has shown the best performance in terms of absolute error for the subsequent short-term carbon emission prediction. This framework not only provides essential technical support for analyzing remote sensing information within the context of carbon neutrality but also introduces innovative insights for carbon emission prediction. • The YOLO framework facilitates subsequent evaluation with valuable data support. • Integration of LSTM and detection results enable forecasting of carbon emissions. • Assessment model provides comprehensive and reliable evaluations of carbon emissions. [ABSTRACT FROM AUTHOR]