Neural network based digit recognition. Still in development.
08/04/16 : Added self made database in data-set/new/
. I wrote 1019 digits by hand (yea, I was determined), the classification rate went up from 94% to 98%. Now it has become really good at guessing my digits, shouldn't be as good with yours (you can still do your own database with learn-mode but pca and normalization have to be done manually until I implement them).
Hold left-button to draw
Click right-button to clear
Hit space to submit to network
Build with SFML
- Create a new digit base (currently comparing your mouse drawings with real handwritten digits..)
- Implement PCA et normalization coef computing
- Learn mode (Adds each drawing to database)
Another application with live camera input github.com/wimacod/Camera-Guess.git
:
- http://www.cs.bham.ac.uk/~jxb/NN/l7.pdf
- http://www.di.unito.it/~cancelli/retineu06_07/FNN.pdf
- ESEO Data Mining Project - M. Feuilloy