Open / Closed Eyes
LBP | HAAR | HOG
Eye detection is not a trivial task, especially if you want to perform it on ARM devices. Before using the following cascades read carefully this page to get the best performance and to know the terms of usage.
The eye detection is a common strategy in computer vision to start the face alignment by Homography or by a more sophisticated method such as Active Shape Models (ASM), Active Appearance Models (AAM), Features Regression, etc. Regardless from the preferred method, a good and a stable initialization of the process is mandatory to get a perfect alignment in a small number of iterations. A perfect alignment usually bring to superior performance in face recognition, identification or classification task. With similar cascades it’s also possible to start detecting the blinking phenomenon for the driving security. Below the HD model for the detection of eyes via boosted cascades.
Human eyes, trained with:
- approx. 9,000 positive samples (randomly sampled)
- approx 0.7 B of negative sub-regions containing faces and non faces samples (90%-10%)
For any other info about Computer Vision and Artificial Intelligence please contact ARGO Vision.
LBP: (contact us)
- Full/partial opened eyes
- Features set: 85.550 features
- Training time: ~1 days
- TP: ~ 95.8% of positive training set
- FN: ~ 04.2% of positive training set
- FP: ~ 7.51937e-006% of negative training set
- Training size w=30 h=60 (aspect ratio 1:2)
LBP: (contact us)
- Full/partial opened and semi/full closed eyes
- Training time: ~2 days
- TP: ~ 97.87% of positive training set
- FN: ~ 02.13% of positive training set
- FP: ~ 9.91927e-006% of negative training set
- Training size w=14 h=28 (aspect ratio 1:2)
OpenCV references: documentation and official guide.