Design Simulation and Analysis of Deep Convolutional Neural Network Based Complex Image Classification System

Abstract

There are 350 families and over 250,000 known varieties of flowering plants. Furthermore, effective flower classification, including content-based image recovery, is essential for the order, plant inspections of buildings, the gardening sector, live plantations, and scientific flower classification guidelines. The representation of flowers has a broad variety of uses. However, manual categorization is time-consuming and exhausting, particularly when the image basis is confusing, has a large number of images, and is perhaps erroneous for several flower groupings. Therefore, effective flower division, discovery, and categorization processes are of great significance. To ensure robust, trustworthy, and ongoing characterization during the preparation stage, new approaches are proposed in this work. On three datasets of flowers that are undeniably known, our technique is tested. Results that are better than the best in this aim for all data sets with accuracy over 98 percent. The categorization of flowers from a wide variety of animal groups is attempted in this research using a unique two-way deep learning method. In order for the foundation box to be placed around the floral area, it was first separated into sections. In a system that uses just convolutional networks, the suggested method for floral distribution is shown to be a parallel classifier. Make a powerful classification using convolutional neural networks in order to recognize the various flower types.