Research And Implementation Of Algorithm For Calculating Correlation Between Image And Text Based On Deep Learning

Recently,there is a growing concern about deep learning(Deep Learning DL)technology which has attracted widespread attention from the industry and the academic.on the one hand,compared with the traditional machine learning model,deep learning has a stronger ability of expression,which can express the complex function more effectively and then achieve higher precision,on the other hand,in the field of computer vision,deep learning model has the ability to automatically abstract features from image which gradually replaces the traditional experience based feature engineering,so that make the technology has a lower threshold.Since 2012,deep learning had brought significant influence in the field of computer vision and natural language processing,but there is still a huge vacancy left in the researching of the cross field these two field.Deep learning still h as tremendous potential for serving the scene which expect computer to handle image and text at the same time.The four main contributes in this paper are as follows:1.We introduce the background and significance of this topic,expound the necessity of using deep learning algorithm,and briefly summariz e the research contents and history of computer vision and natural language processing;2.This topic make an analysis of deep learning models used commonly in computer vision and natural language processing,compare the advantages and disadvantages of them,which providing reference for designment and experiment in this topic;3.Through mining training data with high quality training from log data,we design,implement and optimize the model based on deep learning technology for acquiring the correlation between image and text,and then make a test for this model in the validation set;4.This topic design and propose a model for calculating correlation between image and text finally,which brings convenience to the scene which needs to match the semantics between image and text.Improving image searching engine's ranking ability,which brings a better searching experience for users.