Flower Classification Utilisizing Tensor Processing Unit Mechanism

Abstract

The biodiversity of the species and the potential for visual similarity across the many flower class species, categorizing flowers can be quite a difficult undertaking. The process of classifying flowers is fraught with difficulties, such as blurry, noisy, and poor quality photos, as well as those obscured by plant leaves, stems, and occasionally even insects. With the introduction of deep neural networks, machine learning methods were utilized instead of the conventional handmade features for feature extraction. Because of its quick calculation and efficiency, researchers have shifted their attention to using non-handcrafted features for picture classification tasks. We have discovered several varieties of flowering plants in nature. It is challenging to distinguish and classify the species of flower for education purpose. The identification of objects is expanding across several sectors as a result of the recent development of deep learning in computer vision. In order to get over these issues and constraints, our research created an effective and reliable deep learning flower classifier based on transfer learning and the most advanced convolutional neural networks. According to this study’s suggested model, the Adam optimizer’s accuracy utilising the ResNet50 model is 93 percent.