Fashion MNIST Classifier

A multiclass image classification project built with PyTorch to classify images from the Fashion-MNIST dataset.

Dataset

The project uses the Fashion-MNIST dataset.

• 60,000 training images
• 10,000 test images
• Image size: 28 × 28 pixels
• 10 clothing and footwear classes

Model

A fully connected neural network (MLP) is used for classification.

28 × 28 Image
     ↓
   Flatten
     ↓
 784 → 64
     ↓
   ReLU
     ↓
  64 → 32
     ↓
   ReLU
     ↓
  32 → 10
     ↓
   Logits

The output layer does not use an activation function, as the raw logits are passed directly to CrossEntropyLoss.

Training

• Loss Function: CrossEntropyLoss
• Optimizer: Adam
• Learning Rate: 0.001
• Epochs: 20
• Activation: ReLU

Evaluation

The model is evaluated using Loss, Accuracy, Precision, Recall and F1 Score

Results

Metric	Training	Test
Loss	0.2085	0.3531
Accuracy	92.13%	88.19%
Precision	92.10%	88.37%
Recall	92.13%	88.19%
F1 Score	92.10%	88.20%

Technologies

• Python
• PyTorch
• Torchvision
• TorchMetrics
• Matplotlib
• Jupyter Notebook