click here to download project abstract
ABSTRACT
A fully robotic strategy for recognizing therapeutic plants using PC vision and artificial
intelligence methods has been demonstrated. The leaves of 25 distinctive therapeutic
plant species were collected and captured using an advanced cell in a research
environment. Countless highlights have been separated from each sheet, for example
length, width, border and area, number of vertices, shading, border
and
frame
territory. Then some highlights determined by these properties have been elaborated.
The best results were obtained by a CNN classifier that uses a 10 times cross
approval strategy. With an accuracy of 98.3%, the CNN classifier performed better
than any other AI approach, such as the nearest k- neighbor, Naïve Bayes, KNN and
neural systems. These results are enriching and future work will be equipped to use a
larger data set and elite processing offices to explore the exposure of deep learning
neural systems to recognize restoration plants used in essential medicinal services.
As far as we know, this work is the first of its kind that has created a unique image
data set for therapeutic plants accessible on the island of Mauritius. It is expected that
an electronic or portable PC framework for the programmed recognition of restoration
plants will help the population of the neighborhood to improve their knowledge of
therapeutic plants, to help taxonomists develop recognizable test methods for
increasingly competent species.
