Learn Convolution Neural Networks using TensorFlow, CNN for Image Recognition, and CNN for Object Detection. Understand the concepts and methodologies of CNNs with respect to data science with live coding throughout.
About This Video
Learn Convolution Neural Networks using TensorFlow, CNN for Image Recognition, and CNN for Object Detection. Understand the concepts and methodologies of CNNs with respect to data science with live coding throughout.
About This Video
In Detail
Convolutional Neural Networks (CNNs) are considered game-changers in the field of computer vision, particularly after AlexNet in 2012. They are everywhere now, ranging from audio processing to more advanced reinforcement learning. So, the understanding of CNNs becomes almost inevitable in all fields of data science. With this course, you can take your career to the next level with an expert grip on the concepts and implementations of CNNs in data science.
The course starts with introducing and jotting down the importance of Convolutional Neural Networks (CNNs) in data science. You will then look at some classical computer vision techniques such as image processing and object detection. It will be followed by deep neural networks with topics such as perceptron and multi-layered perceptron. Then, you will move ahead with learning in-depth about CNNs. You will first look at the architecture of a CNN, then gradient descent in CNN, get introduced to TensorFlow, classical CNNs, transfer learning, and a case study with YOLO.
Finally, you will work on two projects: Neural Style Transfer (using TensorFlow-hub) and Face Verification (using VGGFace2).
By the end of this course, you will have understood the methodology of CNNs with data science using real datasets. Apart from this, you will easily be able to relate the concepts and theories in computer vision with CNNs.
Audience
This course is designed for beginners in data science and deep learning. Any individual who wants to learn CNNs with real datasets in data science, learn CNNs along with its implementation in realistic projects, and master their data speak will gain a lot from this course.
No prior knowledge is needed. You start from the basics and slowly build your knowledge of the subject. A willingness to learn and practice is just the prerequisite for this course.
Chapter 1 : Introduction to the Course
Course Overview
Introduction to Instructor
Why CNN
Focus of the Course
Chapter 2 : Image Processing
Gray-Scale Images
Gray-Scale Images Quiz
Gray-Scale Images Solution
RGB Images
RGB Images Quiz
RGB Images Solution
Reading and Showing Images in Python
Reading and Showing Images in Python Quiz
Reading and Showing Images in Python Solution
Converting an Image to Grayscale in Python
Converting an Image to Grayscale in Python Quiz
Converting an Image to Grayscale in Python Solution
Image Formation
Image Formation Quiz
Image Formation Solution
Image Blurring 1
Image Blurring 1 Quiz
Image Blurring 1 Solution
Image Blurring 2
Image Blurring 2 Quiz
Image Blurring 2 Solution
General Image Filtering
Convolution
Edge Detection
Image Sharpening
Implementation of Image Blurring Edge Detection Image Sharpening in Python
Parametric Shape Detection
Image Processing
Image Processing Activity
Image Processing Activity Solution
Chapter 3 : Object Detection
Introduction to Object Detection
Classification Pipeline
Classification Pipeline Quiz
Classification Pipeline Solution
Sliding Window Implementation
Shift Scale Rotation Invariance
Shift Scale Rotation Invariance Exercise
Person Detection
HOG Features
HOG Features Exercise
Hand Engineering Versus CNNs
Object Detection Activity
Chapter 4 : Deep Neural Network Overview
Neuron and Perceptron
DNN Architecture
DNN Architecture Quiz
DNN Architecture Solution
FeedForward FullyConnected MLP
Calculating Number of Weights of DNN
Calculating Number of Weights of DNN Quiz
Calculating Number of Weights of DNN Solution
Number of Neurons Versus Number of Layers
Discriminative Versus Generative Learning
Universal Approximation Theorem
Why Depth
Decision Boundary in DNN
Decision Boundary in DNN Quiz
Decision Boundary in DNN Solution
BiasTerm
BiasTerm Quiz
BiasTerm Solution
Activation Function
Activation Function Quiz
Activation Function Solution
DNN Training Parameters
DNN Training Parameters Quiz
DNN Training Parameters Solution
Gradient Descent
Backpropagation
Training DNN Animation
Weight Initialization
Weight Initialization Quiz
Weight Initialization Solution
Batch MiniBatch Stochastic Gradient Descent
Batch Normalization
Rprop and Momentum
Rprop and Momentum Quiz
Rprop and Momentum Solution
Convergence Animation
DropOut, Early Stopping and Hyperparameters
DropOut, Early Stopping and Hyperparameters Quiz
DropOut, Early Stopping and Hyperparameters Solution
Chapter 5 : Deep Neural Network Architecture
Convolution Revisited
Implementing Convolution in Python Revisited
Why Convolution
Filters Padding Strides
Padding Image
Pooling Tensors
CNN Example
Convolution and Pooling Details
MaxPooling Exercise
NonVectorized Implementations of Conv2d and Pool2d
Deep Neural Network Architecture Activity
Chapter 6 : Gradient Descent in CNNs
Example Setup
Why Derivatives
Why Derivatives Quiz
Why Derivatives Solution
What Is Chain Rule
Applying Chain Rule
Gradients of MaxPooling Layer
Gradients of MaxPooling Layer Quiz
Gradients of MaxPooling Layer Solution
Gradients of Convolutional Layer
Extending to Multiple Filters
Extending to Multiple Layers
Extending to Multiple Layers Quiz
Extending to Multiple Layers Solution
Implementation in NumPy ForwardPass
Implementation in NumPy BackwardPass 1
Implementation in NumPy BackwardPass 2
Implementation in NumPy BackwardPass 3
Implementation in NumPy BackwardPass 4
Implementation in NumPy BackwardPass 5
Gradient Descent in CNNs Activity
Chapter 7 : Introduction to TensorFlow
Introduction to TensorFlow
FashionMNIST Example Plan Neural Network
FashionMNIST Example CNN
Introduction to TensorFlow Activity
Chapter 8 : Classical CNNs
LeNet
LeNet Quiz
LeNet Solution
AlexNet
VGG
InceptionNet
GoogLeNet
Resnet
Classical CNNs Activity
Chapter 9 : Transfer Learning
What Is Transfer learning
Why Transfer Learning
ImageNet Challenge
Practical Tips
Project in TensorFlow
Transfer Learning Activity
Chapter 10 : YOLO
Image Classification Revisited
Sliding Window Object Localization
Sliding Window Efficient Implementation
YOLO Introduction
YOLO Training Data Generation
YOLO Anchor Boxes
YOLO Algorithm
YOLO Non-Maxima Suppression
RCNN
YOLO Activity
Chapter 11 : Face Verification
Problem Setup
Project Implementation
Face Verification Activity
Chapter 12 : Neural Style Transfer
Problem Setup
Implementation TensorFlow Hub
Thank You and Conclusion