Research On Weed Image Recognition In Corn Field Based On GoogLeNet

Corn is one of the most important food crops in China.Farmers favor corn for its wide planting area,high total yield,and agricultural value.However,weeds have become one of the most critical factors affecting the benefit of corn planting.Weed control can effectively reduce weed damage and improve corn quality and planting benefits.The vital technology for weed control is weed identification.Traditional cornfield weed identification mainly relies on artificial eyes,making it difficult to meet practical needs.This paper takes ten common cornfield weeds more harmful in cornfields as the research object.It researches image recognition of cornfield weeds based on Transfer Learning and GoogLeNet network model,mainly completing the following three aspects:Firstly,this study took corn seedlings and ten common weeds in cornfields as objects to establish a cornfield weed dataset.Public datasets websites,web crawlers,and cameras manually captured three ways are used to collect data images.The images are preprocessed,such as manual cleaning,to construct a weed image dataset in multiple ways.Secondly,this paper proposes image recognition of cornfield weeds based on Transfer Learning and the GoogLeNet network model to improve recognition efficiency.The structure of the model is optimized,and the optimized model is compared with the classical neural network.The experimental results show that the model of this study has a tremendous effect on weed identification in the cornfield and can meet the requirements of image recognition of weeds in the cornfield.Finally,weed image recognition software in the cornfield was developed using MATLAB APP Designer in this paper.The software also implements user login,weed control advice,and manual weed functions.After testing,the software designed by this research has strong practicability and stability and can provide specific technical support for the intelligent identification and control of weeds in cornfields and promote the development of agricultural informatization.