Convolutional Neural Networks for the Intuitive Identification of Plant Diseases

Plant phenotyping analysis is in high demand for non-destructive systems, which can accurately calculate plant growth and its structural parameters. In this work the growth of plants was in the indoor conditions. Here the proposed system is a non-destructive for plants using the web camera and plant growth and structured parameters such as leaf area measured with the web camera in the different time periods. We design a growth chamber for the plants in the indoor conditions for that maintained the temperature, humidity and grow lights for to generate the photosynthesis. In the proposed system developed the hardware for the real time plant images using the Arduino and relays. By using the Raspberry pi and web camera capture the images and by using sensors to measure the temperature, humidity and light of the plants. All the images(dataset) will be stored in the cloud. The images will be train and test with the CNN algorithm to classify the whether the plant is healthy or unhealthy. The proposed system attained an average accuracy of 98.2 percent, demonstrating neural network approach’s viability even under adverse conditions.