Upload stage3_fashion_cnn_tuned.h5

---
license: mit
tags:
- image-classification
- cnn
- tensorflow
- keras
- fashion-mnist
- academic-project
datasets:
- fashion_mnist
metrics:
- accuracy
- precision
- recall
- f1
---

# Deep CNN for Fashion MNIST – Tuned Stage 3 Model

This model is part of a staged research project on clothing classification using deep convolutional neural networks and the Fashion MNIST dataset.
It represents **Stage 3**, where a deeper CNN was designed using **Batch Normalisation** and **Dropout** to enhance generalisation and stability.

## 🧠 Architecture

- 3 Conv2D layers: (32, 64, 128 filters)
- Batch Normalisation after each conv layer
- MaxPooling + Dropout
- Fully connected layer (128 units) → Dense(10 softmax)

## 📊 Metrics (Evaluated on Test Set)

| Metric | Value |
|------------|---------|
| Accuracy | 0.9012 |
| Precision | 0.9053 |
| Recall | 0.9012 |
| F1 Score | 0.8992 |

## 🗂️ Dataset

**Fashion MNIST**
- 60,000 training images
- 10,000 test images
- 28×28 grayscale
- 10 clothing categories

## 🧑‍💻 Author

**Alfred Ogunbayo – MSc AI**
[GitHub](https://github.com/freddylags)
[Hugging Face](https://huggingface.co/alfred-ogunbayo)

stage3_fashion_cnn_tuned.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c187d1255d5512a3e01090974cfe4a3e3dcf80b446aeecb883f6ffb272edc75
+size 2986536