Can a probabilistic image annotation system be improved using a co-occurrence approach?
Ainhoa Llorente, Stefan Rüger · 2008
The research challenge that we address in this work is to examine whether a traditional automated annotation system can be im- proved by using external knowledge. Traditional means any machine learn- ing approach together with image analysis techniques. We use as a base- line for our experiments the work done by Yavlinsky et al. (24) who de- ployed non-parametric density estimation. We observe that probabilistic image analysis by itself is not enough to describe the rich semantics of an image. Our hypothesis is that more accurate annotations can be pro- duced by introducing additional knowledge in the form of statistical co- occurrence of terms. This is provided by the context of images that other- wise independent keyword generation would miss. We test our algorithm with two datasets: Corel 5k and ImageCLEF 2008. For the Corel dataset, we obtain statistically significant better results while our algorithm ap- pears in the top quartile of all methods submitted in ImageCLEF 2008. Regarding future work, we intend to apply Semantic Web technologies.