Animal Classification implemented in Farm fields using CNN

Abstract

The day-to-day lives of people depend on the food consumed. Even though food is required regularly, people don’t often think of the struggle the farmers face in delivering the food to the market. There are much more criteria to be considered when it comes to the problems affecting the farmers and the fields. One of the most important criteria is the protection of farm fields. Animal intruding the field leads to crop damage, and of course some severe problems that affect the regular profit. Farm fields near mountain slopes are often intruded by wild elephants and wild pigs, that destroy most of the crops and pull down the profit as well as the investment. There are several old methods to protect the field like thorn fences, but those aren’t quite beneficial. The other problem is the classification of animal entering the field. The security features can be adapted only based on the animal that is entering. If the animal intruder is anonymous, preventive measures cannot be immediately taken. The proposed model uses a setup like fence, where cameras are mounted to capture animal movements using OpenCV python. Once any movement is detected, an alert sound goes on, so that people could be aware that some intrusion has occurred. Using image processing by CNN, classification of animal is done by training and testing the dataset. Precautions along with messages to people who could provide help can be implemented as an additional feature to this proposed work. This structure is considered beneficial to be implemented in military bases to capture movements and alert the soldiers.