To run the demo sucessfully, you must download the model of vgg-face.mat firstly. The download page is here: vgg-face.mat. Put the model of vgg-face.mat to the root path and run the script of extractCNN_VGG_Face.m to extract face feature, then run the script of queryInDatabaseDemo.m to query an image.
I tested the performance on a small face dataset. The search results are as follows:
You may also be interested in a C++ version or a QT version. SeetaFaceLib is a face image retrieval system based on SeetaFaceEngine. I'm very happy that the QT project has been used by someone. The GUI of the QT project is designed as follows:
I have made a video demo. If you inerested in this project and want to know the performace of face image retrieval, you can access to the video: Video Demo