3 ms·

I didn't look at the source in detail, but almost all face detection systems today (including those implemented in hardware in cameras) use a variant of Viola a

by apu 16y ago

I didn't look at the source in detail, but almost all face detection systems today (including those implemented in hardware in cameras) use a variant of Viola and Jones' Haar-Like Cascade Detector [1]. The major exception is work from CMU and PittPatt [2], which uses Neural Networks [3]. [1] http://en.wikipedia.org/wiki/Viola-Jones_object_detection_framework http://en.wikipedia.org/wiki/Viola-Jones_object_detection_fr... [2] http://www.pittpatt.com/ http://www.pittpatt.com/ [3] http://vasc.ri.cmu.edu/NNFaceDetector/ http://vasc.ri.cmu.edu/NNFaceDetector/

liuliu 16y ago
I doubt that PittPatt uses NN. Here is the reference I found: Feature-Centric Evaluation for Efficient Cascaded Object Detection, Henry Schneiderman Maybe Schneiderman himself can clarify this :)
- apu 16y ago
  Ah good point. I forgot that he had his own method, separate from Takeo's.