Hybrid Neural Network Classifier for Detecting Weeds and Plant Diseases in Greenhouses

Abstract

Greenhouse farming plays a crucial role in satisfying the demand throughout the year, regardless of climatic conditions. However, maintaining crop health in greenhouses is critical and challenging. This paper is devoted to analyzing the performance of a hybrid neural network (HNN) based classifier for effective and timely detection of weeds and tomato diseases in greenhouses. Each plant disease has unique features that can be recognized and classified. It is also possible to analyze images of plants in the early stages to detect weeds because weeds are usually significantly different from varietal plants. Thus, we can effectively analyze the condition and type of plants at each stage of cultivation. HNN models can be used to optimize the resource usage for growing a crop unit. It will also allow better monitoring of plant health. In addition, early detection of diseases will significantly reduce the excessive use of agrochemicals.