Bamboo Species classification Using Deep Convolutional Neural Network (DCNN)

Abstract

In Asia, bamboos are the most significant forest product used by rural communities. Different bamboo species are utilized for various purposes. The categorization and identification of bamboo plant species are one of the more challenging tasks in agriculture because of the variety and various field circumstances. The traditional system, which relies on skilled hand labeling, takes a lot of time. Therefore, this study uses deep learning techniques to recognize these kinds of bamboo. In this paper, initially convert the original input image into a gray scale image, then remove the noise from the input images to better predict. Secondly, extracts the features of bamboo culm sheaths such as blade, auricle, ligule, and hairs from the culm sheath using the DenseNet-169 technique. Thirdly, utilizing the YOLOv5 technique to detect the sheaths from bamboo for better classification. And finally, classify the bamboo with its category type using the Deep Convolutional Neural Network (DCNN) technique. For experiments, it collects the input data from the user and the process on our dataset with those collected data, if the input image is matched with our collected data then it will show the category of the bamboo species to the user. This experiment achieves the “state-of-the-art” classification outcomes with 96.17% classification accuracy. The accuracy, sensitivity, specificity, and precision metrics will be utilized with manually gathered data to assess the efficacy of the proposed methodology.