Skip to content

jpthanga/Handwritten_Devangiri_Detector

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Character Detector for Devanagari Text

Devanagari, also called Nagari is an text used in India and Nepal. It is written from left to right, has a strong preference for symmetrical rounded shapes within squared outlines, and is recognisable by a horizontal line that runs along the top of full letters.

Dataset:

Dataset was obtained from: https://archive.ics.uci.edu/ml/datasets/Devanagari+Handwritten+Character+Dataset

Procedure:

Pre-Processing:

The 32 by 32 images are vector converted by flattening into a vector of length 1024.

Dimension Reduction:

The Data was tested using Principal Component Analysis. And a dimension reduction of 16 was decided.

Training:

Training was done on a Feed Forward Neural net with a hidden layer of 50 neurons.

The training accuracy as of now is 60%