Method for identifying and classifying litchi varieties based on spark and deep learning

Abstract

The present application relates to the technical field of automatic identification and classification of litchi varieties, and particularly relates to a litchi variety identification and classification method based on Spark and deep learning, which comprises the following steps: S1. obtaining RGB image data of each variety of litchi and dividing the data into a training set and a test set; S2. constructing a Resnet34_CBAM identification model; S3. training the identification model based on the training set and retaining the identification model with the best performance; S4. building a Spark cluster and a Hadoop cluster, simultaneously providing an AnalyticsZoo, and deploying the optimal identification model in the Spark cluster and the Hadoop cluster through the AnalyticsZoo; and S5. reading in a large amount of litchi images and storing the images in a distributed file system HDFS of the Hadoop cluster, and writing back the images to the HDFS according to categories to realize distributed one-key classification of the large amount of litchi images. The method can obtain a global optimal value more quickly, and makes the convergence effect of the litchi identification model stable, and can quickly improve the operation efficiency of the algorithm.

Images