Is There a Free Lunch for Image Feature Extraction in Web Applications

Abstract

Feature extraction is one of the essential parts of multimedia indexing in similarity search and content-based retrieval methods. Most applications that employ these methods also implement their client side interface using web technologies. The world wide web has become a well-established platform for distributed software and virtually all personal computers, tablets, and smartphones are equipped with a web browser. In the past, most applications employed a strict client-server approach, where the client part (running in the browser) handles only the user interface and the server side handles data storage and business logic. However, the client-side technologies leaped forward with the new HTML5 standard and the web browser has become capable of handling much more complex tasks. In this paper, we propose a model where the multimedia indexing is handled at the client side, which reduces necessary computational power of the server to run a web application that manages large multimedia database. We have implemented an in-browser image feature extractor and compared its performance with a server implementation.