Design And Implementation Of Album Application Based On Semantic Tag Organization And Retrieval Methods

In recent years, with the popularity of mobile devices camera, high-capacity storage and image sharing applications, there are a large amount of personal digital images stored on mobile devices. It becomes hard to manage(organize, retrieve and browse) these mass images without the aid of application. For most image management Apps in App Store, they use the metadata(time, geotag) to organize photo and are lacked of image retrieval, which is hard for user to manage photos in such scale.Aiming at the problem, we proposed "content-based semantic-tagging image organization and retrieval method". We also designed an AI circulatory solution, which helps to improve algorithm performance by collecting users’ feedbacks continuously, and could help extend classes of image recognition and increase accuracy of classification algorithm. Based on the method we proposed, we further implemented an i OS album application, which can automatically sort photos into albums by image content, with image retrieval and album sharing functions.The main work of this paper including:(1) Analysed and compared the different methods of organization and retrieval of i OS Apps and to what extent does these methods meet the needs of users managing a large-scale personal images.(2) Designed and implemented an i OS album application: provides function of image auto-sorting, browsing, sharing, retrieval, wrong-sorting feedback;(3) Ported natural language processing tool word2 vec to i OS platform trying to fill the semantic gap of CBIR; Applied convolutional neural network in the task of image classification.(4) Designed and implemented a server application including: algorithm module, album sharing module and image storage module.In this paper, according to the software engineering development process, the application carried out a detailed analysis, design and implementation. Currently, the application has been initially completed based on the methods we proposed, including i OS App and server application, which could help user to manage growing massive images on mobile devices.