How to become a Computer Vision and AI Guru!

If you're looking at a long career in artificial intelligence, you need to master computer vision and NLP, among other key domains. This article presents a unique CV course in part 1. In part 2, I also share a cognitive AI approach to CV in a case study.

The goal is for an AI agent to acquire the humanlike ability to correlate its environment by visualizing the world, making sense of it with CV, and correlating visual features with concepts.

The first step is a comprehensive CV course.

I. PyImageSearch Gurus Course

I think of this course as "Adrian's world".

Adrian Rosebrock's course is UNIQUE because Adrian is UNIQUE! You can sign up for the course by clicking on the following link.

I have spent a lifetime 24/7 designing AI algorithms at a corporate level including image parsing and optimization. I know what I'm talking about! This course is fantastic.

I found that Adrian, through some kind of miracle, takes you back in history and builds your knowledge up step by step in a unique way to the state of the art of computer vision and AI.

You come into the course as what you are and leave the course as a different person with an Adrian vision(literally) of the world. The world of Adrian!

You'll have a lot of fun with the programs provided! Apply each lesson to something you like or are working on and you will feel like a superhero of computer vision!

I had a lot of fun running the scores of programs and admit I was addicted to reading the course, running the programs, and going through the quiz process!

The following list contains the main topics you need to know, remember or learn about computer vision and that the PyImageSearch Gurus course will take you through. Take a very long and close look at these topics. That's what it takes and you can do it with determination.

CV COURSE TOPICS & PROGRAMS: Loading, displaying, and saving images ?Image basics ?Drawing ?Basic image processing ?Translation ?Rotation ?Resizing ?Flipping ?Cropping ?Image arithmetic ?Bitwise operations ?Masking ?Splitting and merging channels ?Kernels ?Morphological operations ?Smoothing and blurring ?Lighting and color spaces ?Thresholding ?Gradients and edge detection ?Gradients ?Edge detection ?Contours ?Finding and drawing contours ?Simple contour properties ?Advanced contour properties ?Contour approximation ?Sorting contours ?Histograms ?Connected-component labeling Building Your Own Custom Object Detector ?What are object detectors? ?An introduction to object detection ?Template matching ?Object detection The easy way ?How to install dlib ?Object detection made easy ?Sliding windows and image pyramids ?Image pyramids ?Sliding windows ?The -step framework ?Preparing your experiment and training data ?Constructing your HOG descriptor ?The initial training phase ?Non-maxima suppression ?Hard-negative mining ?Re-training and running your classifier ?Training your custom object detector ?Tips on training your own object detectors ?Content-Based Image Retrieval ?What is Content-Based Image Retrieval? ?Your first image search engine ?The steps of building any image search engine ?Defining your image descriptor ?Feature extraction and indexing ?Defining your similarity metric ?Searching ?The bag of (visual) words model ?Extracting keypoints and local invariant descriptors ?Clustering features to form a codebook ?Visualizing words in a codebook ?Vector quantization ?Going from multiple features to a single histogram ?Forming a BOVW ?Inverted indexes and searching ?What is Redis? ?Building an inverted index ?Performing a search ?Evaluation ?Tf-idf weighting ?Spatial verification ?Implementing spatial verification ?Searching with spatial verification ?Evaluating search with spatial verification ?Image Classification and Machine Learning ?A high level overview of image classification ?What is image classification? ?Types of learning ?The image classification pipeline ?k-Nearest Neighbor classification ?Common machine learning algorithms for image classification ?Logistic regression ?Support Vector Machines ?Decision trees ?Random forests ?k-means clustering ?Bag of visual words for classification ?A different type of image pyramid ?Image pyramids for classification ?PBOW ?Image classification example Flowers- ?Image classification example CALTECH- ?Tips on training your own image classifiers ?Face Recognition ?What is face recognition? ?LBPs for face recognition ?The Eigenfaces algorithm ?Preparing and pre-processing your own face data ?The complete face recognition pipeline ?Automatic License Plate Recognition ?What is ANPR? ?The problem with ANPR datasets ?Localizing license plates in images ?Segmenting characters from the license plate ?Scissoring the license plate characters ?Our first try at recognizing license plate characters ?Gathering our own license plate characters ?Improving our license plate classifier ?Tips on classifying your own license plates ?Hadoop + Big Data ?Introduction to Hadoop and MapReduce ?Setting up Hadoop on your machine ?Preparing your images for use on HDFS ?Running computer vision jobs on MapReduce ?High-throughput face detection ?High-throughput feature extraction ?Deep Learning ?Introduction to deep learning ?Neural networks in a nutshell ?Introduction to neural networks ?The Perceptron algorithm ?Multi-layer networks and backpropagation ?Setting up your deep learning development environment ?Deep Belief Networks ?Deep Belief Network basics ?Training a Deep Belief Network ?Convolutional Neural Networks ?A CNN primer ?Training your first CNN ?Implementing CNN architectures ?LeNet ?KarpathyNet ?MiniVGGNet ?Running a pre-trained network ?Transfer learning ?What is transfer learning? ?Transfer learning example dogs and cats ?Transfer learning example flower classification ?Transfer learning example CALTECH- ?Working with Caffe ?Making a dataset compatible with Caffe ?The anatomy of a Caffe project ?Training and evaluating a network with Caffe ?Tips on training your own networks ?Raspberry Pi Projects ?Installing OpenCV on your Raspberry Pi ?Setting up your Raspberry Pi Camera ?Accessing the Raspberry Pi camera and video stream ?Home surveillance and motion detection ?Face recognition for security ?Image Descriptors ?What are image descriptors, feature descriptors, and feature vectors? ?Color channel statistics ?Color histograms ?Hu Moments ?Zernike Moments ?Haralick texture ?Local Binary Patterns ?Histogram of Oriented Gradients ?Understanding local features ?Keypoint detectors ?FAST ?Harris ?GFTT ?DoG ?Fast Hessian ?STAR ?MSER ?Dense ?BRISK ?ORB ?Local invariant descriptors ?SIFT ?RootSIFT ?SURF ?Real-valued feature extraction and matching ?Binary descriptors ?What are binary descriptors? ?BRIEF ?ORB (descriptor) ?BRISK (descriptor) ?FREAK ?Binary feature extraction and matching ?Computer Vision Case Studies ?Measuring distance from the camera to objects in an image ?Face detection in images ?Face detection in video ?Object tracking in video ?Identifying the covers of books ?Plant classification ?Handwriting recognition ?Building Computer Vision Apps for Your Mobile Device ?Introduction to PhoneGap ?Overview of PhoneGap ?PhoneGap environment setup ?PhoneGap “Hello, World” ?PhoneGap UI Setup ?Capturing and uploading a photo with PhoneGap ?Displaying face detection results ?Hand Gesture Recognition ?Introduction to hand gesture recognition ?Hand, finger, and motion segmentation ?Recognizing gestures and more!

?You will be a true state-of-the-art CV Guru by the time you reach the end of the list!

Once you complete the course and pass the quizzes, you will obtain a gratifying PyImageSearch Gurus Certificate.

You can publish the badge on your website or social media profile. It shows that you are CV Guru and that you are keeping up with the market.

II.Cognitive CV-AI applications

I am a strong advocate of adding cognitive AI to stochastic AI. If you are now a CV Guru, you are ready to understand that an AI program can learn features but also concepts.

For example, in Artificial Intelligence by Example, 2nd Edition, 2021, I wrote a chapter on Conceptual Representation Learning(CRL). I trained a CNN to learn how to learn the concept of empty spaces. Empty spaces, or gaps, provide an AI agent with the ability to detect empty parking spaces, the level of production, traffic density, and more. In the same book(chapter 1), I train a reinforcement learning program, an MDP, to optimize itinerary sequences, production sequences, or any kind of sequences.

In Transformers for NLP, 2021, I trained a BERT model and produced a KantaiBERT engine that can learn logic. The following video shows how I assembled the AI Concept Learning Agent, RL, and my KantaiBERT Transformer into a Recommender that can optimize supply chain flows and production.

It requires determination to reach this point of the article by becoming a CV Guru, and understanding the books I mentioned.

To complete the journey, become an NLP Guru as well. I provide a road map in the following article.

CV-NLP-Cognitive AI Guru

If you followed the paths of this article, you are a true Cognitive AI Guru. You can help corporations optimize their critical supply chain path from consumption to production and delivery.

A fabulous new era awaits you!

For more, please ask me questions if you wish here on LinkedIn.