Automatic Detection and Clustering of Actor Faces based on Spectral Clustering Techniques

Abstract

We describe a video indexing system that aims at indexing large video files in relation to the presence of similar faces.

Towards computer-vision software tools to increase production and accessibility of video description for people with vision loss

“…On the technical side, addressing computer-assisted video description requires the development of algorithms that can automatically or semi-automatically extract visual content, time-tag, and organize them according to the needs of the describer [9][10][11][12]. There has been little work in the field of computer-assisted video description.…”

Spectral clustering and semi-supervised learning using evolving similarity graphs

“…In spectral clustering, as in all clustering techniques, nodes that originate from the same cluster should have high similarity values, whereas nodes from different clusters should have low similarity values. Spectral analysis can be applied to a variety of practical problems, including face clustering [11,12], speech analysis [13,14] and dimensionality reduction [15], and, as a result, spectral clustering algorithms have received increasing interest. More clustering applications of spectral graph clustering are reviewed in [16].…”

Spectral face clustering

“…Previous approaches have used different clustering methods like K-Means (K is the user specified number of clusters) and spectral clustering [1] to cluster faces. They use the Euclidean distance metric to represent the distance between faces.…”