https://github.com/heshanera/charrecognizer
Character Recognition using an artificial neural network. Network is trained for 26 characters. Implemented in C++ and Magick++ API is used for pixel manipulation
https://github.com/heshanera/charrecognizer
artificial-neural-networks character-recognition cpp magick-api
Last synced: 12 months ago
JSON representation
Character Recognition using an artificial neural network. Network is trained for 26 characters. Implemented in C++ and Magick++ API is used for pixel manipulation
- Host: GitHub
- URL: https://github.com/heshanera/charrecognizer
- Owner: heshanera
- License: gpl-3.0
- Created: 2017-05-04T18:33:01.000Z (about 9 years ago)
- Default Branch: master
- Last Pushed: 2017-07-27T14:35:28.000Z (almost 9 years ago)
- Last Synced: 2025-02-24T14:47:51.668Z (over 1 year ago)
- Topics: artificial-neural-networks, character-recognition, cpp, magick-api
- Language: C++
- Homepage:
- Size: 109 MB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: license
Awesome Lists containing this project
README
# CharRecognizer
Character Recognition using an artificial neural network. Network is trained for 26 characters. Implemented in C++ and Magick++ API is used for pixel manipulation.
### Neural Network
Network is build with 4 layers, Input layer, Output layer and 2 hidden layers with 80 and 40 nodes. Used Sigmoid and Tangent-Sigmoid functions for the node Activation.
### Training
Network is trained for 26 English uppercase characters from A-Z using 6 training cases for each distinct character.
### Character Recognition
| **Input** |  |  |  |  |  |
| ---------- |-----| -----| -----| -----| -----|
| **Output** | `Recognized: C` | `Recognized: W` | `Recognized: S` | `Recognized: A` | `Recognized: D` |