Improvement of cotton pest image classification algorithm based on MobilenetV3
Cotton is the most popular natural fiber among consumers and is one of the world's most important cash crops.The decline of cotton yield and quality due to various diseases and pests will directly affect the economic benefits of cotton farmers.Tra-ditional image recognition technology is not only time-consuming and laborious,but also easy to make mistakes in judgment.To solve these problems,MobileNetV3,a lightweight network,was selected as the basic model to identify pests and diseases on cotton leaves.Firstly,data enhancement is carried out on the data.Secondly,MobileNetV3 cotton pest image classification algorithm based on transfer learning is proposed to solve the problem that there are few existing cotton pest data sets and the accuracy needs to be improved.Finally,AdamW optimizer was selected for updating,and the appropriate hyperparameters were selected by adjust-ing batch size and learning rate of the model several times.
万方数据
