OCR Algorithms <ul><li>Jacek Bajor </li></ul>
OCR <ul><li>Optical Character Recognition </li></ul><ul><li>Translation of printed or handwritten text to digital form </l...
OCR <ul><li>Fields </li></ul><ul><ul><li>Pattern Recognition </li></ul></ul><ul><ul><li>Artificial Intelligence </li></ul>...
Applications <ul><li>Digitalising libraries </li></ul>
Applications <ul><li>books.google.com </li></ul>
Applications <ul><li>Office work </li></ul>
Applications <ul><li>Captchas </li></ul>Mr. blocked
Applications <ul><li>Recognising licence plates </li></ul><ul><li>And many, many others... </li></ul>
OCR Algorithms <ul><li>Artificial Neural Networks </li></ul>
Why ANN?
Why ANN? <ul><li>They are flexible  </li></ul><ul><li>They can be thought </li></ul><ul><li>They can learn themselves </li...
ANNs <ul><li>Let’s assume ANN is a black magical box </li></ul>
ANNs <ul><li>Learning stage </li></ul>input output a a a a a a
ANNs <ul><li>Evaluation </li></ul>98% a 5% a?
ANN structure Neuron (nerve cell)
ANN structure
ANN structure
ANNs types <ul><li>Feedforward </li></ul><ul><li>Recurrent (back-propagation) </li></ul><ul><li>Radial basis function netw...
ANNs applications <ul><li>Optical Character Recognition </li></ul><ul><li>Function  approximation </li></ul><ul><li>Artifi...
Advantages <ul><li>Flexible </li></ul><ul><li>Efficient if well designed </li></ul>
Disadvantages <ul><li>Focused on one activity </li></ul><ul><li>Give approximations </li></ul><ul><li>Inefficient if poorl...
Libraries <ul><li>FANN - Fast Artificial Neural Network Library - leenissen.dk/fann </li></ul><ul><li>Annie - annie.source...
Thank you
Upcoming SlideShare
Loading in...5
×

OCR

3,648

Published on

Published in: Education, Technology
0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
3,648
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
315
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide

OCR

  1. 1. OCR Algorithms <ul><li>Jacek Bajor </li></ul>
  2. 2. OCR <ul><li>Optical Character Recognition </li></ul><ul><li>Translation of printed or handwritten text to digital form </li></ul>KAZ 813
  3. 3. OCR <ul><li>Fields </li></ul><ul><ul><li>Pattern Recognition </li></ul></ul><ul><ul><li>Artificial Intelligence </li></ul></ul><ul><ul><li>Machine Learning </li></ul></ul><ul><ul><li>Computer Vision </li></ul></ul>
  4. 4. Applications <ul><li>Digitalising libraries </li></ul>
  5. 5. Applications <ul><li>books.google.com </li></ul>
  6. 6. Applications <ul><li>Office work </li></ul>
  7. 7. Applications <ul><li>Captchas </li></ul>Mr. blocked
  8. 8. Applications <ul><li>Recognising licence plates </li></ul><ul><li>And many, many others... </li></ul>
  9. 9. OCR Algorithms <ul><li>Artificial Neural Networks </li></ul>
  10. 10. Why ANN?
  11. 11. Why ANN? <ul><li>They are flexible </li></ul><ul><li>They can be thought </li></ul><ul><li>They can learn themselves </li></ul><ul><li>They are powerful </li></ul>
  12. 12. ANNs <ul><li>Let’s assume ANN is a black magical box </li></ul>
  13. 13. ANNs <ul><li>Learning stage </li></ul>input output a a a a a a
  14. 14. ANNs <ul><li>Evaluation </li></ul>98% a 5% a?
  15. 15. ANN structure Neuron (nerve cell)
  16. 16. ANN structure
  17. 17. ANN structure
  18. 18. ANNs types <ul><li>Feedforward </li></ul><ul><li>Recurrent (back-propagation) </li></ul><ul><li>Radial basis function network </li></ul><ul><li>Kohonen self-organizing network </li></ul><ul><li>Others... </li></ul>
  19. 19. ANNs applications <ul><li>Optical Character Recognition </li></ul><ul><li>Function approximation </li></ul><ul><li>Artificial Intelligence </li></ul><ul><li>Finance </li></ul><ul><li>Speech analysis </li></ul><ul><li>Many, many others </li></ul>
  20. 20. Advantages <ul><li>Flexible </li></ul><ul><li>Efficient if well designed </li></ul>
  21. 21. Disadvantages <ul><li>Focused on one activity </li></ul><ul><li>Give approximations </li></ul><ul><li>Inefficient if poorly designed </li></ul>
  22. 22. Libraries <ul><li>FANN - Fast Artificial Neural Network Library - leenissen.dk/fann </li></ul><ul><li>Annie - annie.sourceforge.net </li></ul><ul><li>Libann - savannah.nongnu.org/projects/libann </li></ul>
  23. 23. Thank you
  1. A particular slide catching your eye?

    Clipping is a handy way to collect important slides you want to go back to later.

×