Mobile NetV2 Maize Seed Variety Recognition Based on Improved Attention Mechanism CBAM
Maize is a main food crop in China,with high nutritional value and economic value.Different geo-graphical environments are applicable for different varieties of maize for planting,however,due to the slight differ-ences in appearance of different varieties,it is difficult to quickly and accurately identify different varieties with the naked eye.In order to realize the accurate identification of maize seed varieties,total 2 792 images of 9 kinds of maize seeds were collected in this study to establish a data set,and training set,verification set and test set were ran-domly divided according to a ratio of 7∶2∶1.In this study,the attention mechanism CBAM was introduced into the lightweight model MobileNetV2,the serial mode of CBAM was improved,a new attention module E_CBAM was built,and different reduction ratios were compared,the best reduction ratio of4 was selected,and E_CBAM_Mobile-NetV2 model was proposed.Experiments indicated,that the accuracy of E_CBAM_MobileNetV2 was 98.18％,5.45％higher than that of MobileNetV2.
NETL
NSTL
万方数据
