Plant Seedlings Classification using Transfer Learning

Abstract

Agriculture is essential for human survival and remains a major economic driver in many countries around the world. Most of the living things around the world feed on vegetation produced by agriculture. Therefore, the researchers should work on developing agriculture using the most recent artificial intelligence approaches. The diagnosis of the plant diseases based on the leaf detection are currently utilized based on machine vision systems. The selective of weeding are more helpful and struggled to identify weeds on a reliable and accurate manner compared with the traditional classification workflows that are sluggish and error-prone results from classification expertise given small number of expert taxonomists. In this paper, an overview of recent attempts to classify species using computer vision and machine learning techniques are realized. It concentrates on identifying plant species using leaf images. We used a dataset containing 4,275 images of 12 species at various growth stages. Furthermore, we present an architecture for plant seedling classification-based machine learning. Convolutional Neural Network (CNN) and transfer learning are utilized as a classification algorithm. The experimentations results based on these classifiers indicated that the proposed model achieved 0.9754, 0.9742, 0.9766, and 0.9754 in terms of Accuracy, Sensitivity, Specificity, and F-score, respectively.