基于改进的Vision Transformer杂草生长周期识别方法研究

Identification of agricultural weed growth cycle based on improved Vision Transformer

通过对农业杂草生长周期的识别分类,达到精准除草的目的。通过使用多卷积模块,使得改进的Vision Transformer模型提取到更加丰富的特征和语义信息,实现对杂草图像生长周期的识别分类。实验结果表明,运用改进Vision Transformer网络模型比传统卷积神经网络识别准确率提升4%。

Through the identification and classification of the growth cycle of agricultural weeds,the purpose of accurate weeding can be achieved.Through the use of multi convolution module,the improved vision transformer model can extract richer feature and semantic information,and realize the recognition and classification of weed image growth cycle.The experimental results show that the recognition accuracy of the improved vision transformer network model is 4%higher than that of the traditional convolutional neural network.