Bangladeshi Local Flower Classification Using CNN

Abstract

A very beautiful gift, known to us as a flower, has been sent by our nature. Flowers have given us all kinds of colors and lovely fragrances as well, and only flowers enhance the beauty of our world. The fragrance of the flower brings immense peace of mind. It mesmerizes everyone. In various parts of our lives, technology will play an important role in helping aspects of our lives. Today's computer vision technology is powered by deep learning algorithms that make sense of images using a special form of the neural network, called a convolutional neural network (CNN). In deep learning, we can use the convolutional neural network (CNN) to get state-of-the-art accuracy in various classification problems, such as image info, CIFAR-100, CIFAR-10, MINIST data sets. In this work, we propose a new system to identify automatic self-ruling decision-making and predictive models using a convolutional neural network for different types of local image flower detections (CNN). A lot of research has been done previously on flower classification in image classification issues, but our related issue of local Bangladeshi flower detection problem does not work on any model and any datasets. We have retrained the final layer of the CNN architecture, MobileNet, Inception V3, VGG16 for classification approach, for solid architecture. Predicting between 6 different types of flower pictures (AKONDO, DADMORDON, DUTURA, KOCHURIPANA, SIALKATA, VATFUL). We suggested an overall accuracy of about 90 percent that can be used for various purposes, such as different implementations of the operating system.