상명대 생물정보학 연구실

Study of alien species monitoring using deep learning

This study aims to rapidly and accurately detect species with the potential to establish populations in Korea using deep learning-based image detection and classification technology. By identifying invasive species at an early stage, we seek to prevent their establishment and effectively manage already introduced species to minimize ecological disruption.

Currently, field surveys conducted by trained specialists are the primary method for monitoring invasive species across large geographic areas. While this approach ensures high accuracy, it is labor-intensive, requires expert personnel, and poses practical challenges in efficiently covering extensive regions with limited resources. To address these limitations, AI-based image recognition technologies, particularly object detection models capable of simultaneous species classification and object identification, have been increasingly applied to wildlife monitoring.

Furthermore, by integrating object detection models with object counting and object tracking technologies, species can be identified and their populations quantified in real time. This study focuses on optimizing these technologies for diverse field applications, enhancing their effectiveness in invasive species monitoring and management.

The deep learning-based invasive species monitoring model developed in this study is expected to significantly contribute to the detection and management of invasive species in Korea, supporting efforts to mitigate ecological disturbances.