Peach Ripeness Grading Method Based on Improved MobileNet V3 Network
At present,in the peach production in China,the peach apparent ripeness is still mainly gra-ded by artificial method based on subjective experience.This method is not only less efficient,but also easily influenced by subjective factors,resulting in uneven maturity grades of the same batch of peaches,which is unable to meet the maturity grade standards required for international sales.To address the above problems,a peach apparent ripeness grading model CS-MobileNet-P-L was put forward in this paper based on the improved convolutional neural network MobileNet V3.Firstly,in order to improve the feature extraction ability of the model,a multidirectional coordinated attention mechanism module was introduced into the original attention mechanism to constitute a dual attention mechanism.Secondly,in order to improve the grading accuracy of the model,the activation function in the Bneck structure of the network was adjusted,and the Last Stage structure of the model was optimized and improved.The results showed that under the same training strategy and envi-ronment configuration,the accuracy of the improved CS-MobileNet-P-L model was 2.71 percentage points higher than that of the MobileNet V3 model,thus better realized the automated and accurate grading of peach apparent ripeness.
PeachApparent ripeness gradingConvolutional neural networkMobileNet V3Attention mechanismActivation function
万方数据
维普
