Transfer Learning Based Effective Approach for Classification of Flowers

Deep learning algorithms are being used to do complex tasks like extracting meaningful features, segmenting, and semantic classification of images. In recent years, these methodologies have had a substantial impact on the classification of flower types. Deep neural networks are a type of image recognition system that has been widely used in computer vision applications because of the similarities between classes and intraclass variance. Due to a greater number of flower species that are similar in shape, color, and visualization, classifying flowers is a tough task. Classification of flowers is used for different purposes, like Recognition of medicinal plants. In this paper, we implemented VGG19 and EfficientNetV2L architecture for the classification of flowers. We have fine-tuned our model compared to other methods to obtain higher accuracy. By normalizing and scaling our flower images, we underwent pre-processing. Then we input the pre-trained model of the images. we divided our flower dataset into train, test, and validation. We achieved an accuracy of 88.21 for VGG19 for 20 epochs and 96.28 for 20 epochs which provides the best accuracy than other proposed architectures in the Kaggle dataset.