Piper Plant Classification using Deep CNN Feature Extraction and Hyperparameter Tuned Random Forest Classification

Abstract

The plant has numerous uses in medicine, food, and industry and plays a major role in environmental protection. Hence it is crucial to identify and classify the specific plant species. In the agriculture production and botanical area, plant classification for images of leaves is considered the basic research. Due to the higher dimensionality and nature complexity of leaf image data, various effective algorithms are required to perform the classification of specific plant species. Hence, in this study, all plant types and specific piper plant types are considered and classified based on the Hyperparameter tuned random forest algorithm due to its effective optimal hyperparameter tuning. Piper plants are selected in this research since they possess significant medicinal applications. Significantly the Deep CNN approach is considered an effective feature extraction of all plants such as tomato, apple, cherry, and others and also piper plants like piper mulesa, piper nigrum, and others. However, initially the effective pre-processing of data augmentation to reduce overfitting and increase the amount of data and feature scaling for data features normalization are established. The experimental results show that the proposed hyperparameter tuned random forest classifier shows better results of showing an accuracy value of 0.94 for all plants and 0.88 value for piper plant compared with other machine learning algorithms like SVM, naïve Bayes, and Logistic regression.

An Improved CNN model for Identifying Tomato Leaf Diseases

“…The average results demonstrate that 73% accuracy may be attained with CNN for the identification of plant pests and diseases in Kenaf plants. [23]…”

Design of an efficient IoT based irrigation platform via fuzzy bioinspired multisensory analysis of on-field parameter sets

“…When processing the information, there must be no instances of lateness, and when there are cases of lateness, the findings must be provided quickly [18]. In addition, the capacity for storage and the rate of processing are very important, but the infrastructure support provided by smart farms is insufficient to meet these requirements [19]. Information that has been detected is sent to platforms for distributed computing, such as the cloud, in order to allow frequent and confined research [20].…”