The National Academy of Marine Research (NAMR) hosted the MDImageNet Indo-Pacific Forum today (October 12) at the IEAT Conference Center in Taipei. The event brought together international experts and research teams from Australia, Korea, Thailand, the United States, and Taiwan to promote the concept of AI-Ready Data Governance, a standardized management framework that ensures image datasets can be directly used for AI model training. The forum emphasized establishing a complete data governance chain—from collection and annotation to sharing and management—to guarantee that marine debris image datasets achieve consistent quality, traceability, and cross-border interoperability. This approach lays a solid foundation for AI applications and fosters regional collaboration in data collection, sharing, and governance across the Indo-Pacific region.
 
The project’s primary goal is to build an AI-Ready Marine Debris Image Dataset that can be accessed by AI teams and research institutions worldwide, rather than directly developing AI models. By creating high-quality, traceable, annotated image data, NAMR aims to enable future AI developers, including students and citizen science participants, to train models on a fair and consistent foundation.
 
International Experts Share Insights on AI and Database Applications
The forum, centered on the AI-Ready Marine Debris Image Dataset, opened with online remarks from NAMR President Chung-Ling Chen. President Chen emphasized that NAMR integrates Taiwan’s technological expertise with the knowledge and experience of its international partners to establish an AI-ready marine debris image dataset. The goal is to develop smarter, faster, and more coordinated tools to address the challenge of marine debris. Guided by the principle of turning data into action, she noted that NAMR’s mission extends beyond Taiwan to the Indo-Pacific and the global stage, striving to build a cleaner, healthier, and more sustainable ocean for future generations.
The forum was moderated by Ning Yen, CEO of IndigoWaters Institute Co., Ltd., who has long been engaged in rapid marine debris screening. Discussions focused on how artificial intelligence and cross-border collaboration can enhance the efficiency and accuracy of marine debris monitoring while promoting data openness and public participation.
 
Experts from across the Indo-Pacific shared their experiences in AI and database applications. Researcher Jian-Wu Lai of NAMR introduced Taiwan’s MDImageNet Marine Debris Image Dataset, developed under the AI-Ready Data Governance framework to ensure consistency and traceability in data collection, labeling, classification, and management. The dataset adopts the marine debris classification system of the International Coastal Cleanup (ICC) and has accumulated over 20,000 images since 2023. The data are publicly accessible through the MDImageNet Marine Debris Image Dataset Platform (https://mdimagenet.namr.gov.tw/), incorporating Convolutional Neural Network (CNN) and object detection technologies for real-time image recognition.
 
Lai noted that marine debris is diverse in form and heavily influenced by environmental factors. Without high-quality, standardized image data, AI applications in monitoring and management remain limited. Therefore, a dedicated dataset is essential to ensure consistent classification and foster cross-national cooperation. In July, NAMR hosted the Marine Debris Survey and MDImageNet Promotion Event at Yehliu, New Taipei City, encouraging the public to photograph debris with their phones and identify it instantly using AI, marking a new model for “tech-powered beach cleanups.”
 
From Thailand, Karun Mooksrisai and Hau Nguyen Cong of the Geoinformatics Center (GIC) at the Asian Institute of Technology presented the pLitter System Database. This innovative platform uses geographic information systems (GIS) and spatial technologies to monitor and manage plastic pollution.
Dr. Sunny Hong, founder of Our Sea of East Asia Network (OSEAN) in Korea, shared the Ocean Knight citizen science initiative and the Ocean Cloud data portal, demonstrating how AI and civic engagement can together build a national-scale marine debris monitoring system.
From the United States, Catherine Prunella of the NOAA Sea Grant Program discussed challenges in marine debris management in the U.S. and proposed new opportunities for community-based datasets and international collaboration.
Dr. Jack Greenshields from Australia’s Tangaroa Blue Foundation introduced the Australian Marine Debris Initiative (AMDI) Database, which compiles data from thousands of cleanups across the country. The database helps identify sources and trends of pollution, driving targeted source reduction and policy reforms, while also exploring future applications of AI in image recognition technology.
 
Turning Data into Action: Advancing Plastic Reduction and Marine Sustainability
In addition to sharing international experiences, the forum also gathered expert feedback on the MDImageNet initiative, laying the groundwork for the upcoming 2026 MDImageNet Image Recognition Challenge. The competition will invite global teams to train next-generation image recognition models using real-world marine debris datasets, encouraging innovative applications that bring solutions into practice.
The forum took place as the second phase of the fifth round of intergovernmental negotiations on the Global Plastics Treaty, convened in Geneva, Switzerland (August 5–14). By deepening exchanges between Taiwan and Indo-Pacific experts, the event allowed participants to gain insight into the latest global trends and best practices in AI and marine debris data management. Together, they reaffirmed a shared commitment to reducing plastic pollution and achieving long-term marine sustainability.