Projects

Handwritten Devanagari OCR using Deep Learning

The Handwritten Devanagari Optical Character Recognition (OCR) project aimed to develop a system capable of recognizing handwritten Devanagari characters. The system utilized a Convolutional Neural Network (CNN) for feature extraction from the handwritten characters and employed OpenCV for image pre-processing. The developed system holds significant potential for diverse applications, including the digitization of handwritten documents and creation of searchable databases for handwritten materials.

Details

Customer Churn Prediction app

A machine learning-powered web application built with Streamlit to predict customer churn based on input features. This tool helps businesses identify customers at risk of leaving, enabling them to take proactive measures and improve customer retention.

🔗 Live App: https://sohel-churn-prediction.streamlit.app/

Details

Spam Email Classifier

Developed a machine learning-based Spam Email Classifier to enhance email filtering efficiency and deployed by making a web-based app. Implemented natural language processing techniques, feature extraction, and a Naive Bayes classification algorithm to accurately identify and classify incoming emails. Achieved high precision and recall rates, contributing to a more effective email communication environment. Technologies employed include Python, scikit-learn, various natural language processing libraries and Flask.

Details