Monthly Archive for August, 2008

Page 2 of 2

Semi-automatic image annotation

Published on 8/11/2008 in References. Closed

L. Wenyin, S. Dumais, Y. Sun, H. Zhang, M. Czerwinski, and B. Field. Semi-automatic image annotation. In In INTERACT2001, 8th IFIP TC.13 Conference on Human-Computer Interaction, pages 326–333. Press, 2001. [PDF]

——

This article describe a technique to incorporate users’ feedback as annotations in a retrieval system for images. The basic argument of the authors is that manual annotation of image might be tedious to the user. Even direct annotation techniques proposed by Sneiderman do not solve the issue. On the other hand automatic annotation of images is not there yet.

Therefore they propose a ‘middle-earth’ approach to the problem asking users of an information retrieval engine to rate the results returned by the system. When results are marked positively, then the system incorporates the query terms as descriptors of the selected images.

To evaluate this architecture, the authors used two metrics: the retrieval accuracy and the annotation coverage. The annotation coverage is the percentage of annotated images in the database. The retrieval accuracy is how often the laveled items are correct. In their experiment, retrieval accuracy is the same as annotation coverage (positive examples are automatically annotated with the query keywords).

The author found that the annotation strategy is more efficient when there are some initial manual annotations. Additionally, they performed a usability test of the system (called MiAlbum) and they found that getting people to discover and use relevance feedback has been difficult. In addition, to improve the discoverability of feedback, the authors argue that we need to improve the participants’ understanding of the matching process.

200808111638.jpg

Does Organisation by Similarity Assist Image Browsing?

Published on 8/3/2008 in References. Closed

K. Rodden, W. Basalaj, D. Sinclair, and K. Wood. Does organisation by similarity assist image browsing? In Proceedings of CHI 2001, Seattle, Wa, USA, March 31-April 4 2001. Association for Computing Machinery. [PDF] [link to author's site]

——-

The title of this paper nicely resumes the authors’ research question. They were interested in understanding whether organizing pictures by similarity might be beneficial to an picture retrieval task. Defining the similarity of two pictures is also an interesting problem and many researcher tried to provide solutions based on the image features or the multi-modal information which might be associated with them. The research reported in this paper was concerned with understanding whether a certain organization might have been more effective in helping the retrieval process than another one.

The authors used two kinds of organizations: 1) similarity of visual features; 2) similarity of text annotations. For the retrieval experiments they used information retrieval’s vector model, with binary term weighting, and the cosine coefficient measure. Also, they used a simulated work task situation, in which they asked graphic designers to look for sets of pictures to be used to complement articles for a magazine.

They conducted two experiments. The first one in which they tried to understand whether text-based organization was more useful than visual-based organization or a combination of the two. The majority of participants favoured the textual arrangements of pictures. In the second experiment, they compared more quantitatively a similarity arrangements to a random arrangement of pictures. They considered the time required to complete the task as the main dependent variable and analyzed the results with a linear regression model. Participants were slower with the visual arrangement than with the random selection of pictures. In the analysis the authors suggested that the visual arrangement made easy to find the target pictures however placing similar pictures together cause sometimes them to appear to merge, and therefore more difficult to parse.

Rodden_CHI_image_matrices.png