Implementation of YOLOv7 for Pest Detection

Abstract

Pests have been known to destroy the yield of the crops, that would soak the nutritional value of the crops. Not only this, but some of the pests can also act as carriers to various diseases that are caused due to the transmutable nature of such bacteria. The most popular pest management technique is pesticide spraying because of how quickly it works and how easily it can be scaled up. Less pesticide use is necessary now, though, as environmental and health awareness grows. Also, existing pest visual segmentation methods are bounding, less effective and time-exhausting, which originates complexity in their marketing and use. Deep learning algorithms have come to be the major techniques to deal with the technological issues linked to pest detection. In this paper, we propose a method for pest detection using a prolific deep learning technique using the newest technology YOLOv7 model. It helps detect which type of pest it is, and if it is a pest that can cause damage, thus by allowing the person to get alert and take appropriate steps. The recommended YOLOv7 model attained the peak accuracy of 93.3% for 50 epochs.