상명대 생물정보학 연구실

A Study on the Classification of Endangered Species by AI-based Image Classification Method

This study aims to quickly and accurately identify endangered species such as parrots and turtles, which are often imported into Korea, by applying AI-based image classification technology.

Image data of internationally endangered species were collected through direct photography and web crawling, and morphological taxonomic features of the studied species were extracted by referring to domestic and foreign papers and books, and image data were selected based on this to build an image database. A bounding box was set using the entire body as the object area to include all the morphological features of the studied species. The image detection training network uses an object detection model that simultaneously performs object area detection and classification, and can identify each object in images containing various species and objects for use in the field of managing the import of internationally endangered species such as watershed agencies and customs offices.

This research is expected to be utilized by Basin Environmental Office, Customs Service, etc. that manage the international trade of endangered species, and it is expected to help conservation of endangered species and prevention of illegal trade.