Introduction:
The objective of this project is to leverage machine learning techniques to classify handwritten Bengali characters. The Bengali script comprises 11 vowels and 39 consonants, totaling 50 basic characters. By employing machine learning algorithms, we aim to develop a robust classification system capable of accurately identifying and categorizing handwritten Bengali characters into their respective vowels and consonants categories. The complexity of handwritten character recognition, particularly in Bengali, poses significant challenges due to variations in size, shape, and individual writing styles. These challenges are increased by similarities in character shapes, and variations in strokes. Bengali, as one of the world's most spoken languages with rich cultural heritage, demands attention in automatic character recognition. Addressing these challenges through machine learning holds promise for enhancing recognition accuracy and advancing linguistic technology. Here we will be using the classical machine learning models to achieve our objective and will try to get as close as possible with the accuracy given by popular deep learning models by incorporating more complex datasets in our study.
What?
- Objective: Develop a model using ML techniques to classify handwritten Bengali characters.
- Bengali Script Details: The script consists of 11 vowels and 39 consonants, making up a total of 50 basic characters.
- Approach: Utilize ML algorithms to identify handwritten characters.
Why?
- Language Significance: Bengali's cultural heritage necessitates effective character recognition solutions.
- Current Limitations: Recognition systems for Bengali lag behind than those available for other languages.
- Challenges: Variations in size, shape, and writing styles complicate recognition process.
How?
- Literature Review
- Data Collection
- Data Cleaning and Preprocessing
- Applying ML Algorithms and Assessment of Performance
- Preparation Of Project Report
Team Members
- Dipankar Dey
- Saikat Kumar Ghosh
Project Information:
Dataset Used:
- Ekush Bangla Handwritten Dataset.
- Bangla Lekha Isolated Dataset.
- Our Collected Primary Dataset.