Pest Recognition System Based On Deep Learning

Forestry not only has the ecological effect,but also has huge economic effect.In recent years,the damage of scolytidae is more serious,resulting in the death of a number of pine trees.The traditional recognition methods rely on manpower,time-consuming and laborious.So accurate and intelligent recognition of the pests in scolytidae is very important.The specimens collected from the Forestry College of Beijing Forestry University were selected,and the data set of the pests in scolytidae was made by indoor picture automatic collection device,and it was divided into training set and test set.Based on the TensorFlow framework,we trained the Faster R-CNN model through training set data under Python environment to achieve the purpose of recognizing the pests in scolytidae,and verified its accuracy through test set.Experiments show that the average accuracy rate of six pests in scolytidae is high,and the model can meet the requirements of practical research.By using this model,we design and develop the pest recognition system based on deep learning,which provides a new basis for forestry pest recognition.