This app is based on face detection, recognition and analysis. PornstarID is a
Python and Torch implementation of face recognition with deep neural networks (DNN).
The following overview shows the workflow for an example image:
1. Detect faces with dlib/OpenCV
2. Transform the face for the neural network. PornstarID uses dlib's real-time pose estimation with OpenCV's affine transformation to try to make the eyes and bottom lip appear in the same location on each image.
3. Use a deep neural network to represent (or embed) the face on a 128-dimensional unit hypersphere. The embedding is a generic representation for anybody's face. Unlike other face representations, this embedding has the nice property that a larger distance between two face embeddings means that the faces are likely not of the same person.
Those properties makes clustering, similarity detection, and classification tasks easier than other face recognition techniques where the Euclidean distance between features is not meaningful.
Our algorithm used is trained on 7,008 adult performers and the current neural network was trained on more than 1 million faces in total. Our bots are spidering millions of additional images and videos each month.
Below are a few aligned and cropped faces of a pornstar:
With the next release (version 1.1, scheduled for summer 2017) our neural network will feature over 20,000 pornstars and tens of thousands of webcam girls. In the meantime you can search the database of all the pornstars the DNN is currently trained on here.
Interesting articles about PornstarID: