BOTANICAL AI FOR PLANT IDENTIFICATION AND CONSERVATION

Abstract

This paper explores the possibility of Artificial Intelligence (AI) in species plant identification and the implication to biodiversity protection.  We trained and enhanced deep learning algorithms, particularly convolutional neural networks (CNNs), to automatically and accurately distinguish plant species based on a huge collection of labelled images of plants.  Its classification rates are on average more than 95 % climbing to near 100% making the system better than the traditional taxonomic methods of identification in speed and scalability.  Species recognition was part of the model along with geospatial metadata; thus, spatial distribution could be mapped, and habitats could be described.  The findings revealed that the approach could identify rare and endangered species which aided in facilitating strategies of conservation that would work best.  It was demonstrated in field tests that AI predictions are robust even amid alterations in the environment, whether they are greenhouse controls to a broad variety of natural conditions.  Moreover, integrating AI and mobile-based citizen science applications facilitated the involvement of more people in plant monitoring through the assistance of the latter which contributed to a higher amount of biodiversity data.  The result of this study is that plant identification using AI is a game changer in conservation biology because it enables a quicker, more accurate, and more scalable measurement of biodiversity.  This approach bridges the gap between technology and ecology to motivate us to take an active interest in conservation and therefore simplifying the task of sustaining plant diversity in the rapidly changing world.