himanshurawlani / practical_intro_to_tf2

Building an image classifier in TF2

Practical Introduction to TensorFlow 2.0

This repository introduces the new Tensorflow 2.0 in a practical way by building an image classifier which classifies 5 classes of flowers. It covers the following:

1. Downloading and preprocessing data using TensorFlow Datasets

• Checking out available datasets and their features
• Downloading the dataset (tfds.load()))
• Pre-processing the dataset
• Visualizing the dataset

2. Building and training an image classifier model using Keras high level API

• Building a simple CNN in Keras
• Visualising the model
• Compiling and training the model
• Training the model using data augmentation
• Using TensorBoard inside notebooks

3. Downloading and fine-tuning InceptionV3 pre-trained model

• Downloading pre-trained model
• Adding classification head
• Training the classification head
• Fine tuning the model

4. Serving the trained model using TensorFlow Serving

• Tensorflow Serving installation
• Starting TensorFlow Serving
• Making REST requests
• Parsing the response

Read the Medium article explaining the above concepts in detail here.