THIS is the real Todo List:
- Do maths to work out gaze direction!!!
- work out wether to use COM or local min to improve eeye estimate
- Make the eye template stick to the face
- Track eyes better within the eye template
- Use eye movement to track graze location
- Use template location to get head location and find gaze location
- make a few bitmap/videos and try tracking eye movement, just use simple no head movement first.
- make find maxima etc to I can properly test all the methods.
- Finish working out how to find maxima…
- WordPress about maxima finding attempts->need to find them first
- Test if Gauss func should be a circle or an oval “base”. – perhaps call it a cross section??
- Investigate Joelet size… if I ever get round to it :p
- Learn some fuzzy logic?? Could replace Gaussian blur? Or help to weight estimates?
- Install FFTW.
- Read through latest papers and add to literature review… need Alex to check my current first.
- read more c++ book 😀
- Work out curvature of paper1… sqrt(Ixx^2 + 2Ixy^2 + Iyy^2) – this is just larger for smaller circles, smaller for larger circles. so smaller circles would have more effect. wordpress about this.
- Make/check Gauss func has as constant area underneath it, make A(sigx, sigy) or something. – might be impossible, or need a creative solution. or not relevant, as the image is then just drawn as its fraction from 1-255, so only seems darker, as all points are near the same value, apart from perhaps one or two. maybe should invert image when using high values??
- OpenCV muthaaaa