Topic Title:

Natural Language Based Personal Photo Retrieval

 

Technical Area:

Computer Vision, Natural Language Processing

 

Background

Personal photos have increased dramatically due to the proliferation of smart phones and digital cameras. How to manage and retrieval relevant photos from a large-scale personal photo collection becomes a practical and important issue. Traditional text-query based image retrieval largely relies on the offline generated tags for each photo in the album, which is not effective if the user’s text query does not match the pre-defined tags.

 

Target

Thus, this project calls for solutions that can allow a user to freely use natural language to communicate with the photo search engine to find the relevant photos that he wants. Specifically, we expect the solution is able to: 1) allow the user to use a phrase, or a sentence as query; 2) retrieval the relevant photos in a fast manner.

 

Related Research Topics

Image retrieval, natural language processing, feature embedding, deep learning.