In this project we performed a content-based image retrieval: Generation of an ordered list of images related to a query image. This model searches for the most similar face of a famous person with respect to a query face.
We trained the models on faces contained in CelebA dataset (~200000 images of ~10000 actors).
More details about the processing and architecture in Slides_Image-Retrieval.pdf
. Dimostrative video and deployed model in the DEMO
folder.