🐶 End-to-End Multi-Class Dog Breed Classification

This project is an end-to-end multi-class image classifier designed to identify the breed of a dog from an image using TensorFlow 2.0 and TensorFlow Hub.

1. Problem

Given an image of a dog, the goal is to identify its breed from a set of 120 possible breeds. For example, when you take a photo of a dog at a cafe, the model can predict its breed.

2. Data

The dataset is sourced from Kaggle's Dog Breed Identification Competition. It contains:

• 10,000+ labeled training images (120 breeds).
• 10,000+ unlabeled test images for prediction.

3. Evaluation

The model's performance is evaluated using a file containing prediction probabilities for each dog breed for each test image. Details of the evaluation metric can be found here.

4. Features

• Leverages deep learning and transfer learning for image classification.
• 120 distinct dog breeds (120 output classes).
• Dataset contains 10,000+ images for training and 10,000+ images for testing.
• Built with TensorFlow 2.0 and TensorFlow Hub for state-of-the-art performance.