Vecrist/mnist-pytorch-digit-classifiers

MNIST Handwritten Digit Classifiers (CNN & MLP PyTorch Models)

This repository contains pre-trained PyTorch model weights for classifying handwritten digits (0 to 9) from 28x28 grayscale images:

• CNN Model (MNIST_CNNmodel_weights.pth): 99.03% Test Accuracy.
• MLP Model (MNIST_MLPmodel_weights.pth): Baseline Multi-Layer Perceptron.
• Application: Interactive Tkinter digit drawing app (handdrawnDigitClassification.py).

---

Model Architectures

1. Convolutional Neural Network (CNN) - 99.03% Accuracy

• Block 1: Conv2d(1, 32, kernel_size=3) -> BatchNorm -> ReLU -> MaxPool2d(2)
• Block 2: Conv2d(32, 64, kernel_size=3) -> BatchNorm -> ReLU -> MaxPool2d(2) -> Dropout2d(0.25)
• Classifier: Flatten -> Linear(64*7*7, 128) -> ReLU -> Dropout(0.5) -> Linear(128, 10)

2. Multi-Layer Perceptron (MLP)

• Flatten -> Linear(784, 128) -> ReLU -> Linear(128, 64) -> ReLU -> Linear(64, 10)

---

Model Usage

import torch
import torch.nn as nn
from huggingface_hub import hf_hub_download

# Define CNN Model
class MNISTCNN(nn.Module):
    def __init__(self):
        super().__init__()
        self.features = nn.Sequential(
            nn.Conv2d(1, 32, 3), nn.BatchNorm2d(32), nn.ReLU(), nn.MaxPool2d(2),
            nn.Conv2d(32, 64, 3), nn.BatchNorm2d(64), nn.ReLU(), nn.MaxPool2d(2), nn.Dropout2d(0.25)
        )
        self.classifier = nn.Sequential(
            nn.Flatten(),
            nn.Linear(64 * 7 * 7, 128), nn.ReLU(), nn.Dropout(0.5),
            nn.Linear(128, 10)
        )
    def forward(self, x):
        return self.classifier(self.features(x))

model = MNISTCNN()
weights_path = hf_hub_download(repo_id="Vecrist/mnist-pytorch-digit-classifiers", filename="MNIST_CNNmodel_weights.pth")
model.load_state_dict(torch.load(weights_path, map_location="cpu"))
model.eval()