I would almost forget, but I also presented some work at the FG2008 conference: Acceptability Ratings by Humans and Automatic Gesture Recognition for Variations in Sign Productions.
Abstract: In this study we compare human and machine acceptability judgments for extreme variations in sign productions. We gathered acceptability judgments of 26 signers and scores of three different Automatic Gesture Recognition (AGR) algorithms that could potentially be used for automatic acceptability judgments, in which case the correlation between human ratings and AGR scores may serve as an ‘acceptability performance’ measure. We found high human-human correlations, high AGR-AGR correlations, but low human-AGR correlations. Furthermore, in a comparison between acceptability and classification performance of the different AGR methods, classification performance was found to be an unreliable predictor of acceptability performance.
Snapshots of the three signs used in the experiment.
Examples of three manipulations of the sign SAW. We tested about 68 sign manipulations in total. These were run through the automatic recognition algorithms we had been working on and they were rated by human signers. The paper is about how humans and machines can be compared.