Abstract
This report presents a scalable method for automatically discovering frequent visual objects in large image collections even if their size is very small. It extends the work initially published in [12] with additional experiments comparing the proposed method to the popular Geometric Min-hashing method. The basic idea of our approach is that the collision frequencies obtained with hashing-based methods can actually be converted into a prior probability density function given as input to a weighted adaptive sampling algorithm. This allows for an evaluation of any hashing scheme effectiveness in a more generalized way, and a comparison with other priors. In this work, we introduce a new hashing strategy, working first at the visual level, and then at the geometric level. It allows integrating weak geometric constraints into the hashing phase and not only neighborhood constraints as in previous works. Experiments show that this strategy boosts the performances considerably and clearly outperforms the state-of-the-art Geometric Min-Hashing method.