Vlado Menkovski bio photo

Vlado Menkovski

Assistant Professor at Eindhoven University of Technology.

Email Google Scholar

Currently a lot of image recognition models are at hand pre-trained. They are good at either determining what is being shown on an image, what person is there if we know it’s a person and what can we learn from a person (gender, age) knowing that it’s a specific type of person. None of them however manage to do the whole pipeline where given an image we able to determine roughly what is being shown on the image (animal, building, human) and if a specific class is present (human in our case), to next determine traits that are of interest such as gender, age, race. Additionally, if the person is a human, can we determine if this person is present in a pre-defined set of people known beforehand (is the human a celebrity or not)?