CNN Based Animal Repelling Device for Crop Protection

Abstract

The use of modern IoT technology in agricultural regions has been the subject of much research and several attempts. This paper’s main goal is to protect the crop against animal assaults. Traditional methods apply the same level of security to all sorts of animals identified using a Passive IR sensor, and only single-stage protection is used. With the aid of Support Vector Machine and Convolution Neural Network algorithms, the photos were properly recorded and recognized, and the information was delivered to the farm owner via IoT devices. A section of the farm was used to create the project. On either side of the entry, cameras were installed to record images for processing to identify the animals, and different levels of security were applied based on the animal identification. The dB level of the reciprocating sound will fluctuate according to the animal. Different levels of protection and different forms of protection are used depending on the classification of the animals. Making noise and lighting from the opposite side sends the animal out of the farm in the first degree of protection. The collected photographs are sent to the owner at the second level. The suggested method’s accuracy may be determined by comparing it to the standard technique’s complexity, implementation cost, reciprocating time, and animal detection accuracy.