find your favourite pornstar in real-time!
Posted: Sun Sep 18, 2016 6:43 pmSome of you might find the website http://pornstar.id/ useful and thats why I am posting it here.
PornstarID is a new real-time pornstar identification app which was launched September 15th for the public. The current version is beta with only 596 models which the algorithm is able to detect. Over 218,000 images where used in order to train the DNN for those 596 models. PornstarID is able to recognize the correct model close to 50% of all submitted images as number one result. With each result PornstarID is displaying up to 10 possible matches of the recognized face. The correct result (if known to the algorithm) over 90% of the times within those alternative models. Those values are expected to increase with the final launch of PornstarID. The final version will be able to recognize over 20,000 adult performers and is currently in development.
PornstarID is a Python and Torch implementation of face recognition with deep neural networks (DNN). The workflow for a an input image is first to detect faces with dlib. Next pornstarID transforms 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. In the next step PornstarID uses 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.
List of current supported adult performers: http://pornstar.id/adult_performers.php
FAQ: http://pornstar.id/faq.php
Twitter: https://twitter.com/pornstarid
PornstarID is a new real-time pornstar identification app which was launched September 15th for the public. The current version is beta with only 596 models which the algorithm is able to detect. Over 218,000 images where used in order to train the DNN for those 596 models. PornstarID is able to recognize the correct model close to 50% of all submitted images as number one result. With each result PornstarID is displaying up to 10 possible matches of the recognized face. The correct result (if known to the algorithm) over 90% of the times within those alternative models. Those values are expected to increase with the final launch of PornstarID. The final version will be able to recognize over 20,000 adult performers and is currently in development.
PornstarID is a Python and Torch implementation of face recognition with deep neural networks (DNN). The workflow for a an input image is first to detect faces with dlib. Next pornstarID transforms 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. In the next step PornstarID uses 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.
List of current supported adult performers: http://pornstar.id/adult_performers.php
FAQ: http://pornstar.id/faq.php
Twitter: https://twitter.com/pornstarid