Computer Vision Bootcamp with Python (OpenCV) – YOLO, SSD

Introduction
1 Introduction

Convolutional Neural Networks (CNNs) Based Approaches
2 The standard convolutional neural network (CNN) way
3 Region proposals and convolutional neural networks (CNNs)
4 Detecting bounding boxes with regression
5 What is the Fast R-CNN model
6 What is the Faster R-CNN model
7 Original academic research articles

You Only Look Once (YOLO) Algorithm Theory
8 What is the YOLO approach
9 YOLO algorithm – grid cells
10 YOLO algorithm – intersection over union
11 How to train the YOLO algorithm
12 YOLO algorithm – loss function
13 YOLO algorithm – non-max suppression
14 Why to use the so-called anchor boxes
15 Original academic research article

You Only Look Once (YOLO) Algorithm Implementation
16 YOLO algorithm implementation I
17 YOLO algorithm implementation II
18 YOLO algorithm implementation III
19 YOLO algorithm implementation IV
20 YOLO algorithm implementation V
21 YOLO algorithm implementation VI
22 YOLO algorithm implementation VII

Single-Shot MultiBox Detector (SSD) Theory
23 What is the SSD algorithm
24 Basic concept behind SSD algorithm (architecture)
25 Bounding boxes and anchor boxes
26 Feature maps and convolution layers
27 Hard negative mining during training
28 Regularization (data augmentation) and non-max suppression during training
29 Original academic research article

SSD Algorithm Implementation
30 SSD implementation I
31 SSD implementation II
32 SSD implementation III
33 SSD implementation IV
34 SSD implementation V

Appendix #1 – Neural Networks Theory
35 Artificial neural networks – inspiration
36 Backpropagation explained
37 Applications of neural networks I – character recognition
38 Applications of neural networks II – stock market forecast
39 Types of neural networks
40 Artificial neural networks – layers
41 Artificial neural networks – the model
42 Why to use activation functions
43 Neural networks – the big picture
44 Using bias nodes in the neural network
45 How to measure the error of the network
46 Optimization with gradient descent
47 Gradient descent with backpropagation

Appendix #2 – Deep Neural Networks Theory
48 Deep neural networks
49 Activation functions revisited
50 Loss functions
51 Gradient descent stochastic gradient descent
52 Hyperparameters

Appendix #3 – Convolutional Neural Networks (CNNs)
53 Convolutional neural networks basics
54 Feature selection
55 Convolutional neural networks – kernel
56 Convolutional neural networks – kernel II
57 Convolutional neural networks – pooling
58 Convolutional neural networks – flattening
59 Convolutional neural networks – illustration

Appendix #4 – Support Vector Machines (SVMs)
60 What are Support Vector Machines (SVMs)
61 Linearly separable problems
62 Non-linearly separable problems
63 Kernel functions

COURSE MATERIALS (DOWNLOADS)
64 Download source code
65 Download slides

Environment Setup
66 Installing Python and PyCharm on Mac
67 Installing OpenCV
68 Installing Python and PyCharm on Windows

History of Computer Vision
69 Evolution of computer vision related algorithms

Handling Images and Pixels
70 Images and pixel intensities
71 Handling pixel intensities I
72 Handling pixel intensities II
73 Why convolution is so important in image processing
74 Image processing – blur operation
75 Image processing – edge detection kernel
76 Image processing – sharpen operation

Computer Vision Project I – Lane Detection Problem (Self-Driving Cars)
77 Lane detection – the problem
78 Lane detection – handling videos
79 Lane detection – first transformations
80 What is Canny edge detection
81 Getting the useful region of the image – masking
82 Detecting lines – what is Hough transformation
83 Hough transformation illustration
84 Drawing lines on video frames
85 Testing lane detection algorithm

Viola-Jones Face Detection Algorithm Theory
86 Face detection problem introduction
87 Viola-Jones algorithm
88 Haar-features
89 Integral images
90 Boosting in computer vision
91 Cascading
92 Original academic research articles

Face Detection with Viola-Jones Method Implementation
93 Face detection implementation I – CascadeClassifier
94 Face detection implementation II – CascadeClassifier parameters
95 Face detection implementation III – tuning the parameters
96 Face detection implementation IV – detecting faces real-time

Histogram of Oriented Gradients (HOG) Algorithm Theory
97 Histogram of oriented gradients basics
98 Histogram of oriented gradients – gradient kernel
99 Histogram of oriented gradients – magnitude and angle
100 Histogram of oriented gradients – normalization
101 Histogram of oriented gradients – big picture
102 Original academic research article

Histogram of Oriented Gradients (HOG) Implementation
103 Showing the HOG features programatically
104 Face detection with HOG implementation I
105 Face detection with HOG implementation II
106 Face detection with HOG implementation III
107 Face detection with HOG implementation IV