Citrus Huanglongbing identification method and device based on Mixup algorithm

The invention discloses a citrus Huanglongbing identification method and device based on a Mixup algorithm, and the method comprises the steps: carrying out the preprocessing of a to-be-classified image, and carrying out the data enhancement of the preprocessed image; carrying out Mixup algorithm pr...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: JIANG SHENG, HUANG ZHONGQIANG, WANG WEIXING, YAO GANGDONG, LAI JUNGUI, LI WANGZHI
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention discloses a citrus Huanglongbing identification method and device based on a Mixup algorithm, and the method comprises the steps: carrying out the preprocessing of a to-be-classified image, and carrying out the data enhancement of the preprocessed image; carrying out Mixup algorithm processing on the image after data enhancement, and establishing a linear relation between samples; carrying out bN batch normalization processing on the pixel points of the processed image, so that the pixel points accord with normal distribution; and inputting the normally distributed pixels into adepth separable volume integral model based on model fusion, and outputting a classification result of the image. According to the method, the limitation that a traditional data enhancement method depends too much on an original image and is insufficient in generalization of a resistance sample is solved, and the depth separable convolution model based on model fusion is adopted, so that the recognition efficiency is impr