Giant jellyfish blooms are widely recognized as a major marine biological hazard, causing damage to fisheries, safety incidents at beaches, and disruption of marine ecosystems. In recent years, the frequency of mass jellyfish occurrences has increased compared to the past due to climate change and shifting ocean environmental conditions, with notable changes in the timing of appearance and the extent of their distribution. This has heightened the need to quantitatively assess the scale of jellyfish outbreaks and their transport pathways. While conventional visual surveys and citizen-reported sightings are useful for rapidly obtaining occurrence locations, they are limited in systematically characterizing outbreak scale and movement pathways due to inconsistent survey coverage, irregular spatial sampling, and the absence of standardized quantification criteria.

To address these limitations, this study applies a deep learning–based object detection model together with a Lagrangian particle-tracking technique to estimate the abundance and inflow pathways of giant jellyfish entering Korean waters. Shipboard observations were conducted at key survey stations along the inflow period of jellyfish transported via the Kuroshio Current into Korean waters. Field data were collected using custom-built visual observation equipment and drones. Based on imagery acquired at East China Sea survey stations and the Ieodo Ocean Research Station in May and July 2025, automatic jellyfish detection was performed using the You Only Look Once (YOLO) deep learning object detection model. The model achieved a mean Average Precision (mAP) of 0.863 and a maximum F1-score of 0.85, confirming high detection performance. Using bounding boxes generated by the trained YOLO model, jellyfish body size and swimming speed were further estimated. Additional survey data to be collected in May and July 2026 will be incorporated to expand the training dataset and improve the model's generalization performance.

Following detection, jellyfish transport pathways were simulated using a Lagrangian particle-tracking approach based on satellite altimetry–derived surface current data provided by the Copernicus Marine Environment Monitoring Service (CMEMS), enabling identification of sea regions with a high likelihood of mass jellyfish occurrence. By integrating remote sensing–based automated monitoring of jellyfish imagery with post-observation pathway tracking, this study provides a more quantitative understanding of giant jellyfish outbreak scale and dispersal pathways. The approach is expected to support the jellyfish warning and crisis alert system operated by the National Institute of Fisheries Science (NIFS), and to contribute to the future development of a predictive framework for mass jellyfish occurrence.