Скачать или смотреть MNIST Handwritten Digit Classification | CNN Model Deployment using Flask (Demo)

MNIST Handwritten Digit Classification | CNN Project - Team Kernels

This video demonstrates the deployment of a Convolutional Neural Network (CNN) model trained on the MNIST Handwritten Digit Recognition dataset as part of the EN3150 Pattern Recognition module assignment.

The project explores the use of different optimizers and deep learning techniques for image classification, and this demo specifically showcases the Adam optimizer version of the trained CNN model, deployed on Render using Flask.

🌐 Live Deployment

👉 Try it here: https://mnist-flask-app-ef8w.onrender...
🧩 GitHub Repository (Project): https://github.com/akindu-k/cnn-proje...
🧩 GitHub Repository (Deployment): https://github.com/akindu-k/mnist-fla...

📘 Project Overview

This CNN-based project was developed as part of Assignment 03 for EN3150 – Pattern Recognition at the University of Moratuwa.
The objective was to design, train, and evaluate CNN architectures for digit classification, while analyzing the impact of optimizers and transfer learning.

🔹 Implemented Models

CNN with SGD Optimizer
CNN with SGD + Momentum
CNN with Adam Optimizer ✅ (Deployed version)
Transfer Learning using pretrained networks (VGG16, ResNet18)

⚙️ Technologies Used

Python, Flask, PyTorch
NumPy, Pandas, Matplotlib, Scikit-learn
Deployed on: Render

👨‍💻 Team Kernels

Bandara A.H.M.D.T - 220057C
Kalhan M.K.A. - 220298N
Perera I.A.S. - 220464V
Wijewardena L.T.N. - 220728K


#CNN #MNIST #DeepLearning #PatternRecognition #PyTorch #Flask #MachineLearning #UniversityOfMoratuwa #TeamKernels #AI #ComputerVision #EN3150