Water Level detection in Farm Surveillance System using Machine Learning Algorithm

Abstract

Video surveillance system along with IP camera is fast growing led for detecting events from the captured videos. Video Surveillance system in farm is useful for detecting water level. Using Farm Surveillance System, water level is detected automatically and this data is useful in predicting how much water need to supply in the farm. Fourier Transform and Gaussian Low pass lter techniques are used for detecting water level from the captured videos. A NARX neural network is used along with data of last one month to train neural network and to predict water level of a farm. The accuracy of this method is measured based on Mean Squared Error (MSE) and Regression (R) values. The simulation result demonstrate that NARX Time series Neural Network can be used to forecast water level of a farm.