Orangebark: The simplest AI model to discriminate tree species from bark images

• 발행기관: (사)한국식물생명공학회
• 저자명: Narae Kim Chayanee Chairattanawat Jaewook Kim
• 간행물 정보: 『Journal of Plant Biotechnology』52권 4호, 279~289쪽, 전체 11쪽
• 주제분류: 농수해양 > 기타농수해양
• 파일형태: PDF
• 발행일자: 2025.12.31

국문 초록

Research on artificial intelligence (AI)-assisted discrimination of tree species using simple images is ongoing. Evaluation of the performance of almost all such models developed to date poses a significant technical hurdle. Thus, there is an urgent need for an easy-to-use AI model or pipeline for AI model development. Using the Orange3 visual programming environment, we developed an AI model called Orangebark, which is capable of identifying tree species from RGB images of their bark. To develop Orangebark, we utilized two independent public image datasets comprising thousands of images. We constructed a systematic pipeline to examine each step involved in constructing AI models using Orange3. Using this pipeline, we developed Orangebark, which can distinguish 18 tree species with an area under the ROC curve greater than 0.99 and a recall value of 0.873. The pipeline will enable plant biology researchers to readily optimize the parameters for development of AI models, thereby enhancing the pace of AI model development, and provides an easy-to-use model for discrimination of tree species. We anticipate that Orangebark will be widely utilized owing to its extreme ease of use.

목차

Introduction

Materials and Methods

Results

Discussion

References

키워드

Neural Network Bark Tree phenomics