How to perform face recognition with K-Medoids Clustering

[Fender96]

Hi all,
I have a problem doing face recognition with K-Medoids Clustering. I have read the Image Data in KNIME with the Image Reader Knot. The Dataset is a bunch of black and white pictures of faces with several emotions. I have to perfom K-Medoids Clustering on the pictures for face recognition. Is is it the right way to split every picture in vectors? At the moment i involved the Image Properties Node but it does not work out. So is there any kind of Knot to this or is there another solution to solve the Problem?

Thanks for helping out!

[stelfrich]

Hi @Fender96,

Welcome to the KNIME Community!

@Fender96

The Dataset is a bunch of black and white pictures of faces with several emotions. I have to perfom K-Medoids Clustering on the pictures for face recognition. Is is it the right way to split every picture in vectors? At the moment i involved the Image Properties Node but it does not work out. So is there any kind of Knot to this or is there another solution to solve the Problem?

Have you, by chance, looked at the manuscript at K-MEDOIDS CLUSTERING USING PARTITIONING AROUND MEDOIDS FOR PERFORMING FACE RECOGNITION? It’s the first manuscript that an internet search for “face recognition with k-medoids” reveals, which also describes the detailed process. Among the first steps is preprocessing, i.e. extraction of features from individual images, which is required for clustering on the extracted features.

Best,
Stefan