p319-vonahn - CHI 2004 Paper 24-29 April Vienna Austria...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
Labeling Images with a Computer Game Luis von Ahn and Laura Dabbish School of Computer Science Carnegie Mellon University Pittsburgh, PA, USA {biglou,[email protected] Abstract We introduce a new interactive system: a game that is fun and can be used to create valuable output. When people play the game they help determine the contents of images by providing meaningful labels for them. If the game is played as much as popular online games, we estimate that most images on the Web can be labeled in a few months. Having proper labels associated with each image on the Web would allow for more accurate image search, improve the accessibility of sites (by providing descriptions of images to visually impaired individuals), and help users block inappropriate images. Our system makes a significant contribution because of its valuable output and because of the way it addresses the image-labeling problem. Rather than using computer vision techniques, which don’t work well enough, we encourage people to do the work by taking advantage of their desire to be entertained. I.2.6 [ Learning ]: Knowledge acquisition. H.3.m [ Information Retrieval ]: miscellaneous. H.5.3 [ HCI ]: Web-based interaction. General Terms: Design, Human Factors, Languages Keywords: Distributed knowledge acquisition, image labeling, online games, World Wide Web. INTRODUCTION Images on the Web present a major technological challenge. There are millions of them, there are no guidelines about providing appropriate textual descriptions for them, and computer vision hasn’t yet produced a program that can determine their contents in a widely useful way. However, accurate descriptions of images are required by several applications like image search engines and accessibility programs for the visually impaired. Current techniques to categorize images for these applications are insufficient in many ways, mostly because they assume that the contents of images on the Web are related to the text appearing in the page. This is insufficient because the text adjacent to the images is often scarce, and can be misleading or hard to process [4]. The only method currently available for obtaining precise image descriptions is manual labeling, which is tedious and thus extremely costly. But, what if people labeled images without realizing they were doing so? What if the experience was enjoyable? In this paper we introduce a new interactive system in the form of a game with a unique property: the people who play the game label images for us. The labels generated by our game can be useful for a variety of applications. For accessibility purposes, visually impaired individuals surfing the Web need textual descriptions of images to be read aloud. For computer vision research, large databases of labeled images are needed as training sets for machine learning algorithms. For image search over the Web and inappropriate (e.g., pornographic) content filtering, proper labels could dramatically increase the accuracy of
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 02/24/2010 for the course COMM 4400 at Cornell.

Page1 / 8

p319-vonahn - CHI 2004 Paper 24-29 April Vienna Austria...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online