Image classification and annotation based on Hadoop cloud computing platform

Image classification and annotation based on Hadoop cloud computing platform

Lu Huang Hou Li Bin

In order to deal with the massive image and video data of the Internet, a new image classification and annotation solution based on Hadoop cloud platform is proposed. Aiming at how to efficiently extract the training set, a cloud-based image grabbing platform is built, and the image resources of the Internet are used as the original data set. Providing sufficient data for extracting training set images; It realizes the training set image extraction function based on probabilistic latent semantic analysis model, carries on the topic clustering to the original dataset, helps the user to select the training set image quickly, joins the SVM classification model, and uses the training set to classify the image without annotation, The complete system is implemented. The experimental results show that the proposed scheme can meet the function and performance requirement of mass image data classification and annotation.

Image classification and annotation based on Hadoop cloud computing platform