Automated Plant Identification Through Deep Learning with Particular Focus On Medicinal Plants

Ecology, pharmacology, agriculture, and conservation require species identification. Identification is difficult when there are several identical species. Flora-with’s 400,000 species need skilled identification. Dichotomous keys might take days to identify a species, especially in biodiversity hotspots. Non-scientists may struggle with this identifying process. Manual plant identification is difficult and slow, hence an automated system that extracts descriptive elements from raw visual plant data to identify and output the plant species is needed. The initiative suggests using leaves to identify plants instead than flowers, seeds, bark, or seeds. The project aims to create cloud-based item recognition software with a computer vision-based search engine to identify plants faster. Deep learning, a machine learning branch, has excelled in image recognition and classification. Deep learning is crucial to automated medicinal plant identification. Traditional medicine and botany require medicinal plant identification. Manual identification is slow and error-prone. Deep learning-based medicinal plant identification solves these problems. Deep learning algorithms can spot patterns and make predictions from enormous data sets by learning sophisticated relationships between input properties and output labels. Deep learning algorithms can examine leaf shape, texture, and colour to accurately identify plants from a large dataset of plant pictures. Deep learning models are scalable and may be readily integrated into mobile and internet applications, making automated medicinal plant identification solutions user-friendly and accessible. Deep learning in automated medicinal plant identification is a promising and growing field that could transform plant identification and traditional medicine research. This dissertation uses machine learning/deep learning (convolutional neural networks) to recognize plants in 2D pictures.