Deep-Learning-Image-Classification

A neural network built with TensorFlow/Keras to classify images of clothing from the Fashion-MNIST dataset.

Image Classification with a Neural Network

Project Overview

This project introduces the fundamentals of deep learning and computer vision by building a neural network to classify images of clothing. The model is trained on the Fashion-MNIST dataset, a collection of 70,000 grayscale images across 10 different fashion categories. The goal is to demonstrate the end-to-end process of building, training, and evaluating a neural network using modern tools like TensorFlow and Keras.

---

Dataset

The project uses the Fashion-MNIST dataset, which is conveniently included with the Keras library. It is split into 60,000 images for training and 10,000 images for testing. Each image is a 28x28 grayscale representation of a piece of clothing.

---

Methodology

The workflow for this project involved the following key steps:

1. Data Loading & Preprocessing: The data was loaded directly from Keras. The primary preprocessing step was normalizing the pixel values from a 0-255 range to a 0-1 range to optimize the training process.

2. Model Architecture: A sequential neural network was designed using the Keras API with the following structure:

   • An Input Layer (Flatten) to convert each 28x28 image into a 1D vector.
   • A Hidden Dense Layer with 128 neurons and a ReLU activation function to learn the underlying patterns in the images.
   • An Output Dense Layer with 10 neurons (one for each class) and a softmax activation function to produce a probability distribution for the final classification.

3. Training & Evaluation: The model was compiled using the adam optimizer and sparse_categorical_crossentropy loss function. It was then trained for 10 epochs, and its performance was evaluated on the unseen test set to measure its real-world accuracy.

---

Results

The trained neural network achieved a test accuracy of approximately 88%. This result demonstrates a solid ability to build and train a functional computer vision model from scratch. The training history showed a consistent increase in accuracy and decrease in loss, indicating a successful learning process.

This project is a foundational piece for tackling more complex computer vision tasks.

---

Technologies Used

• Python 3
• TensorFlow & Keras: For building and training the neural network.
• NumPy: For numerical operations.
• Matplotlib: For visualizing the data and training results.

---

➡️ Explore the details of the project