tajammulbasheer / deeplizard-keras

This repository contains the notebooks, I created while learning Keras from DeepLizard. It contains cat vs dog classification notebook and sign language clasification notebook.

Keras with Tensorflow by DeepLizard

Project Description

Keras with Tensorflow is a course offered by Deep Lizard on YouTube with freeCodeCamp These are notebooks I wrote following this course.
Note: Anything good and helpful in these notbooks should be credited to freeCodeCamp.org and DeepLizard. Anything foolish or incorrect assume it as my error.
This basic course only uses the upper level features of Tensorflow api with help of Keras

Contents

• 0. Prerequisites
• 1. Sequential Model
• 2. CNN
• 3. VGG16 and Cat vs Dog
• 4. MobileNet
• 5. Mobilenet and Sign Language Digits
• 6. Data Augementation

Prerequisites

• Describing course prequitistes.
• Introducing Keras and Tensorflow
• How to setup GPU and CUDA

Sequential Model

• Data Preprocessing for Neural Networks
• Creating an Artificial Neural Network with Keras
• Trainimg a Simple ANN
• Validation of model with Validation Set
• Making predictions using the model
• Saving and loading model

CNN

• Image data preperation for CNN
• Creating a simple CNN for image classification of Cat vs Dogs
• Building and training the CNN
• Making predictions using CNN

VGG16

• Introducing concept of functional API
• Converting VGG16 to sequential model
• Fine tuning VGG16 for Cat Dog Classification
• Training fine tuned model
• Making Predictions using fine tuned model

MobileNet

• Introduction MobileNet
• Preparing images fore MobileNet
• Making predictions directly on images using MobileNet

MobileNet and Sign Language Digits

• Preparing data for MobileNet
• Modify the MobileNet for Sign Language Digits
• Fine tune the MobileNet for image classification

Data Augementation

• Learnt about augementing image data in different ways

Technologies Used

• Python
• skit-learn
• Keras

Requirments

• Python 3.x
• TensorFlow 2.0 or higher
• scikit-learn
• Keras

Installation

1. Clone the repository
2. Install the required libraries using pip install
3. Download the dataset from below
4. Run the Jupyter notebook to train and test the different models

How to use

1. How to Use
2. Open the Jupyter notebook.
3. Load the data by specifying the path to the dataset.
4. Train the models using the different techniques mentioned in the notebook.

Resources

Papers

Very Deep Convolutional Networks for Large-Scale Image Recognition
MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications

Datasets

Cat vs Dog on Kaggle here
Sign Language Digits Images on Git here

Others

Keras also has it's own beautiful website, providing useful information about how things are to be used with written code examples. That can be found here

Contributions

This project is open to contributions, feel free to fork the repository and make pull requests, and it whatever way you want.