Greenhouse Plant Growth Supervision with the LED Lights using Machine Learning

Abstract

Agriculture is considered the main resource of global economic growth. Greenhouses are glass buildings used to provide plants with their special needs in climate and growth. Our proposed approach tends to establish an automated greenhouse control system for speeding up the plant growth and increasing their production. Controlling the greenhouse is established using Arduino, real-time cameras, LED lights and fans. Different types of sensors are used such as DHT11, soil moisture and LDR. A web application is developed to monitor and track the greenhouse's parameters and the plants' growth. Masking with Hue-Saturation-Value (HSV) is used to detect the desired green range of the plant and the desired color range of the fruit/vegetable. Features are extracted using Histogram of Oriented Gradients (HOG) algorithm and One-Class Support Vector Machine (OC-SVM) as a classifier to detect the fruit/vegetable in the image. The proposed approach achieved 81.8% accuracy in the tomato's classification.