Pest Identification Using Deep Learning

Pest Identification Using Deep Learning

Sayali Govilkar1, Rutuja Jadhav2, Shubhada Randive3, Sanjyot Thuse4

1Department of E&TC PES Modern College of Engineering, Pune, India

2Department of E&TC PES Modern College of Engineering, Pune, India

3Department of E&TC PES Modern College of Engineering, Pune, India

4Department of E&TC PES Modern College of Engineering, Pune, India

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - Pest infestation is indeed a major problem faced by farmers, and it can lead to significant damage to crops. In order to combat this issue, farmers often use pesticides asa means of pest control. Due to less awareness about pesticide and its quantity; lot of experiments are carried out on crops. The excessive and improper use of pesticides can have serious consequences on the environment, wildlife, plants, and human health. It causes various health disorders such as breathing problem, skin cancer and many more. In order to overcome this problem framers should know which pest is harming the crop so that they can use pesticides made for that pest. Our system focuses on pest attacking and harming grapes. Seven classes of insects like Climbing cutworm, Yellow jacket, Grape flea beetle, Multi colored Asian lady beetle, Leaf hopper, Grape phylloxera , Grape berry moth. The system uses deep learning algorithm for Identification of pests. CNN along with YOLOV3 is used for pest identification. An accuracy of 98% is achieved in pest identification. We have developed web application using Django framework where we get by default IP address. The system is developed in the form of web application where the farmer can upload an image, and the pest isidentified immediately with accuracy.

Keywords— Agriculture, CNN, YOLOV3