# Alphabet-Recognition-EMNIST

**Repository Path**: ice_cream_hg/Alphabet-Recognition-EMNIST

## Basic Information

- **Project Name**: Alphabet-Recognition-EMNIST
- **Description**: Alphabet recognition using EMNIST dataset.
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-05-17
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Alphabet Recognition [![](https://img.shields.io/github/license/sourcerer-io/hall-of-fame.svg?colorB=ff0000)](https://github.com/akshaybahadur21/Alphabet-Recognition-EMNIST/blob/master/LICENSE.txt)  [![](https://img.shields.io/badge/Akshay-Bahadur-brightgreen.svg?colorB=ff0000)](https://akshaybahadur.com)
This code helps you classify different alphabets using softmax regression (lower case).

### Sourcerer
[![](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/images/0)](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/links/0)[![](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/images/1)](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/links/1)[![](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/images/2)](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/links/2)[![](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/images/3)](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/links/3)[![](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/images/4)](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/links/4)[![](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/images/5)](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/links/5)[![](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/images/6)](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/links/6)[![](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/images/7)](https://sourcerer.io/fame/akshaybahadur21/akshaybahadur21/Alphabet-Recognition-EMNIST/links/7)

### Code Requirements
You can install Conda for python which resolves all the dependencies for machine learning.

### Description
Softmax Regression (synonyms: Multinomial Logistic, Maximum Entropy Classifier, or just Multi-class Logistic Regression) is a generalization of logistic regression that we can use for multi-class classification (under the assumption that the classes are mutually exclusive). In contrast, we use the (standard) Logistic Regression model in binary classification tasks.

For more information, [see](https://www.kdnuggets.com/2016/07/softmax-regression-related-logistic-regression.html)


### Python  Implementation

1) Dataset- Extended MNIST dataset (letters)
2) Images of size 28 X 28
3) Classify alphabets from a to z
4) Logistic Regression, Shallow Network and Deep Network Support added.

### Train Acuracy ~ 91 to 99%
### Test Acuracy ~ 70 to 84%

<img src="https://github.com/akshaybahadur21/Alphabet-Recognition-EMNIST/blob/master/emnist.gif">

### Execution for writing through webcam
To run the code, type `python Alpha-Rec.py`

```
python Alpha-Rec.py
```