Get Started

 

Now, this is the tutorial that I wish I had access to when I was starting my journey learning OpenCV, in this tutorial I have revealed some very interesting info about OpenCV including excellent tips for you, regarding where to find the right resources, tutorials for the library. The outline:

• How to navigate the OpenCV docs to find what you’re looking for.
• How to get details regarding any OpenCV function.
• The differences between the C++ and Python versions of OpenCV and which one you should work with.
• Pip installation of OpenCV vs Source installation.
• Where to ask questions regarding OpenCV when you’re stuck.

 

In this tutorial, you will learn how to install OpenCV from Source in Windows 10 with Nvidia GPU Support & Non-Free Flags Enabled. The outline of the tutorial, I only recommend that you follow this tutorial if you’re already comfortable with OpenCV and need to utilize additional features that are not available with the default pip installation.

• Pros and Cons of Installation with Package Manager (pip/conda)
• Pros and Cons of Installation from Source
• Step-by-step Installation Procedure.
• Using GPUs with OpenCV DNN Module

 

If you only had time to learn 1 OpenCV algorithm then that should definitely be contours, using contours, you’ll be able to create systems like vehicle detection, object counters, shape detection, distance measurement, etc. without even using deep learning. So it’s a must-learn technique in OpenCV.  So in this tutorial, you will learn the basics of contour detection, how to use it, and go over various preprocessing techniques required for Contour Detection. Below is the outline of the tutorial.

• • • What is a Contour?
    • Contour Detection
    • Visualizing the Contours detected
    • Pre-processing images For Contour Detection
    • Drawing a selected Contour
    • Retrieval Modes

Detecting and drawing contours alone won’t be enough to build powerful computer vision applications. The contours need to be processed and manipulated to make them useful for building cool applications. 

So in this tutorial, you will learn how to perform different contour manipulations to extract beneficial information from the detected contours. Using this information you can easily build very useful applications, in fact, we have an entire course that will help you master contours for building computer vision applications. Here’s the outline of this tutorial.

• • • Detect and draw Contours
    • Extracting the Largest Contour in the image
    • Sorting Contours in terms of size
    • Drawing a rectangle around the contour
    • Drawing Convex Hull

In this tutorial, you will be learning about analyzing contours that help recognize the object being detected and differentiate one contour from another. We will also explore how you can identify different properties of contours to retrieve useful information. After learning to analyze the contours, you will be able to do all sorts of cool things with them. The outline:

• • • Detect and draw Contours
    • Image Moments
    • Find the center of a contour
    • Finding Contour Area
    • Contour Properties

 

Now that you have learned the basics of contour detection, it is time to combine those concepts and start building cool real-time vision applications and learn to use contours at an expert level utilizing this extremely affordable course. 

In the course, we have provided high-quality graded quizzes, applications, case studies, challenges, forums, and a lot more with no AI or Computer Vision background required.

The course we mentioned above is full of exciting real-world applications if you are still confused about whether you should join the course or not. Here is a free tutorial, in which you can see how useful and simple it is to manipulate contour & build real-world applications like vehicle detection. The outline of the tutorial:

• Background Subtraction
• Noise Removal
• Contour Detection & Manipulation

 

Now that you have learned the basics of OpenCV and Image processing, your next step should be to join our premium course on Image Processing with OpenCV ( with video lectures in Urdu/Hindi). The course covers:

• Image Processing
• Classical Computer Vision
• Computer Vision Projects
• Machine Learning & Deep Learning

This course gives you everything you need, in detail for you to master Computer Vision and Image Processing with OpenCV in python with no prerequisites other than just some programming experience in any language, so I highly recommend embracing the opportunity and don’t miss it out.

Nice going! you have completed the first and most important step of your learning journey so you should be proud of yourself. Now the next step is to move towards building some interesting applications and extend your OpenCV knowledge further.

Note: If you have joined and completed my courses then the next section will be a piece of cake for you.

Computer Vision & Machine Learning in the security industry, coupled with sensors like CCTV cameras, etc is a highly demanded skill these days and allows you to derive meaningful information and insights from visual data without anyone’s involvement. This opens up the door of feasibility to a ton of interesting applications like Automatic Number Plate Recognition (ANPR), Anomaly & Violation detection, and  Social Distancing Monitoring.

This tutorial will let you get started in the security industry and guide you in building a robust Intruder Detection surveillance system using your phone’s camera. The system will record video samples whenever someone enters your room and will also send you alert messages via Twilio API. The tutorial can be split into 4 parts:

• • • Accessing the Live stream from your phone to OpenCV.
    • Learning how to use the Twilio API to send Alert messages.
    • Building a Motion Detector with Background Subtraction and Contour detection.
    • Making the Final Intruder Detection System Application

In this tutorial, you will learn to build your own Virtual Pen & also a Virtual Eraser without needing any hardware other than just a normal webcam and a pen. So you will be able to draw and erase virtually on the screen by waving a pen in the air, cool right? So let’s get into it. The outline:

• • • Find the color range of the target object and save it.
    • Apply the correct morphological operations to reduce noise in the video
    • Detect and track the colored object with contour detection.
    • Find the object’s x,y location coordinates to draw on the screen.
    • Add a Wiper functionality to wipe off the whole screen.
    • Add an Eraser Functionality to erase parts of the drawing.

 

 

It’s important to have fun every once in a while, so how about taking a break to play some games. If you want you can go play with your friends or maybe how about we build an AI to play games with us? Sounds like a plan, right? This way you will end up having fun along with learning several new AI concepts at the same time. 

So this tutorial will cover creating a system that will allow you to play Rock, Paper, Scissors with AI. The system will be capable of recognizing your hand signs and then play its random move, as the game does not require a strategy to choose the move. The tutorial can be breakdown into the following steps:

• • • Step 1: Gather Data, for rock, paper scissor classes.
    • Step 2: (Optional) Visualize the Data.
    • Step 3: Preprocess Data and Split it.
    • Step 4: Prepare Our Model for Transfer Learning.
    • Step 5: Train Our Model.
    • Step 6: Check our Accuracy, Loss graphs & save the model.
    • Step 7: Test on Live Webcam Feed.
    • Step 8: Create the Final Application.

Well done! You have completed the section and learned to build some very interesting applications that have added great value to your portfolio, so congrats on that.

 

Image segmentation is one of the most important and widely used image processing techniques as it brings down the classification task to the pixel level and partitions an image into multiple segments (i.e., set of pixels) which makes it much easier to extract and analyze the objects of our interest and be used to make a ton of vision applications.

In the tutorial, you will learn to perform Real-Time Selfie Segmentation using Mediapipe in Python and then build the following 4 applications:

• • • Background Removal/Replacement
    • Background Blur
    • Background Desaturation
    • Convert Image to Transparent PNG

The same segmentation model is also being utilized by Google Hangouts. You will not need any GPU, all of these applications will run smoothly in real-time on your CPU, and trust me there is no catch here. And you will also learn about the following image segmentation categories:

• • • Semantic Segmentation.
    • Instance Segmentation
    • Panoptic Segmentation
    • Saliency Detection.

Pose detection is a must learn technique in vision since it has a lot of real-world use cases like controlling HCI applications, Fitness/exercise/dance monitoring, and overlaying virtual clothes or other accessories over the body, etc.

 

This tutorial will teach you to perform 3D pose detection and pose classification on images, videos, and webcam feed in real-time. Plus at the end I’ll also show you how to create a pose classification system for detecting Yoga poses. The tutorial is divided into the following parts:

• • • Part 1 (a): Introduction to Pose Detection
    • Part 1 (b): Mediapipe’s Pose Detection Implementation
    • Part 2: Using Pose Detection in images and on videos
    • Part 3: Pose Classification with Angle Heuristics

You have made it this far which tells me that you are quite dedicated to becoming a Computer Vision/Machine Learning expert and will succeed soon but one should also maintain a healthy balance between working and having fun.

Regularly doing physical exercises is a great way to stay healthy. So why not combine exercising, coding, and even gaming? And hit all the birds with a single shot.

For this purpose, I have made a tutorial in which you will learn to control a popular game called “Subway Surfers” using body gestures and movements while exercising and having fun all at the same time, with the help of Mediapipe’s pose detection solution utilizing just a webcam. 

Here’s a Demo of me controlling the game using the application we will cover in this tutorial:

The application is divided into smaller components and all the steps to create the components and integrate them to produce the final application are described in a very neat manner. The outline:

• • • Step 1: Perform Pose Detection
    • Step 2: Control Starting Mechanism
    • Step 3: Control Horizontal Movements
    • Step 4: Control Vertical Movements
    • Step 5: Control Keyboard and Mouse with PyautoGUI
    • Step 6: Build the Final Application

 

Hand landmark detection is one of the most popular computer vision techniques out there and every vision practitioner must learn it to compete in the market. Some of its applications are sign language recognition and controlling HCI applications with hand gestures, etc.

This tutorial will teach you to perform 3D hands landmarks detection, and hands classification (i.e. is it a left or right hand) images and videos as well in real-time using the Mediapipe library in python on your CPU with 30+ FPS. 

The tutorial will also cover drawing bounding boxes around hands and customized landmarks annotation of the hands utilizing the Depth of the landmarks.  The tutorial is divided into the following parts:

 

• • • Part 1 (a): Introduction to Hands Recognition & Landmarks Detection
    • Part 1 (b): Mediapipe’s Hands Landmarks Detection Implementation
    • Part 2: Using Hands Landmarks Detection on images and videos
    • Part 3: Hands Classification (i.e., Left or Right)
    • Part 4 (a): Draw Bounding Boxes around the Hands
    • Part 4 (b): Draw Customized Landmarks Annotation

Gesture Recognition is gonna be one of the highest demand skills in the near future as it is a building block of Augmented Reality (AR) applications and can also be used in controlling simple HCI applications via hand gestures which kind of sounds much cooler than typing through keys or tapping on a touch screen.

So in this tutorial, you will learn to utilize the hand landmarks to count hand fingers, build a Hand Gesture Recognizer that will be capable of identifying multiple gestures, and create a Selfie-Capturing System that will be controlled using hand gestures. The outline:

• • • Step 1: Perform Hands Landmarks Detection
    • Step 2: Build the Fingers Counter
    • Step 3: Visualize the Counted Fingers
    • Step 4: Build the Hand Gesture Recognizer
    • Step 5: Build a Selfie-Capturing System controlled by Hand Gestures

With the completion of this section, you have obtained a fine grip on Mediapipe. If you want you can check out the other solutions provided by Mediapipe here. 

In this tutorial, you will learn about the various important details of the DNN module, that are never discussed like, selecting preprocessing params correctly and designing pre and post-processing pipelines for different models, and a lot more that is required to use the module to its fullest extent. This tutorial can be split into 3 sections.

• • • Introduction to OpenCV’s DNN module.
    • Using a Caffe DenseNet121 model for classification.
    • Important Details regarding the DNN module, e.g. where to get models, how to configure them, etc.

In this tutorial, we will first train a custom image classifier with Tensorflow’s Keras API, and use it in the OpenCV DNN module. You can skip the training part for now and focus on learning how to use a custom-trained classifier in OpenCV along with its advantages. You will learn to:

• • • Convert a Classifier to ONNX format
    • Use the ONNX model in the OpenCV DNN module

This tutorial first covers a complete end-to-end pipeline for training a custom object detector and then how to use it in the OpenCV DNN module. For this one too, you can skip the training part for now and focus on learning how to use a custom-trained object detector in OpenCV. You will learn to: 

• • • Export Graph.pbxt with frozen inference graph.
    • Use the Frozen inference graph along with the Pbtxt file in OpenCV.

 

You may already know that the higher the resolution (i.e., number of pixels) is, the more detailed the image is, so the more information it contains to extract. 

Super Resolution is a very useful algorithm that is effectively being used to generate high-resolution versions of the images in various applications (like medical Imaging, Surveillance Systems, and Satellite Imagery, etc) across several industries and is a must-learn technique for every computer vision practitioner.

So you must check this tutorial, which covers some theoretical details of what exactly Super-Resolution is, its algorithms and architectures, and how to perform it with just OpenCV. The outline:

• • • What is Super-Resolution
    • Image Super-Resolution Using Deep Convolutional Networks (SRCNN)
    • Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network (SRResNet)
    • Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network (SRGAN)
    • Enhanced Super-Resolution Generative Adversarial Networks  (ESRGAN)
    • Other Areas Of Super Resolution
    • Implementation of  Super-Resolution with OpenCV 

In this tutorial, you will learn to perform Super-Resolution using multiple models, even those that will do 8x resolution. In the previous tutorial, we had implemented only a single SR model with 3x Resolution, so this one’s gonna be quite interesting. The four different models that you will be learning to use in this tutorial are:

• • • EDSR: Enhanced Deep Residual Network 
    • ESPCN: Efficient Subpixel Convolutional Network 
    • FSRCNN: Fast Super-Resolution Convolutional Neural Networks
    • LapSRN: Laplacian Pyramid Super-Resolution Network 

To achieve effective interaction between humans and machines it is essential for machines to recognize the emotions of humans, as emotions play a vital role in conveying non-verbal messages which can’t be ignored. And emotion recognition can also be used to create other intelligent systems like Smart Music players, Student Mood Monitoring System, and Smart Advertisement Banners, etc.

This tutorial will cover how to perform Facial Expression Recognition AKA Emotion Recognition using the DNN module. The tutorial is structured in the following way:

• • • Emotion Recognition & its Importance.
    • Different approaches to tackle this problem.
    • Implementation of an Emotion Recognition pipeline using OpenCV’s DNN module. 

Nice going! You have completed one more section and learned everything you need to know to effectively implement the Deep Neural Network module in OpenCV.

This tutorial covers the TensorFlow 2.0 GPU installation in a step-by-step manner structured in a very neat and simplest way possible. And not only that it will also help you to set up your Nvidia GPU for OpenCV source installation so that you can use your GPU with the OpenCV DNN module. The outline:

 

• • • Installing Tensorflow 2.0 GPU in Windows
    • Setting Up Your Nvidia GPU for OpenCV DNN 

In this tutorial, we will train an image classifier with Tensorflow’s Keras API, and use it in the OpenCV DNN module. You have already learned to utilize the OpenCV DNN  module so you should only focus on learning how to train a custom classifier with TensorFlow.

 

In this tutorial, you will see a complete end-to-end pipeline for training an object detector and then use it in the OpenCV DNN module. For this one too, you have already done the OpenCV DNN part so you can focus on learning how to train an object detector with TFOD API. You will learn to:

• • • Setup the Environment.
    • Install & Setup TFOD API.
    • Collect & Annotate Data.
    • Download Model & Configure it.
    • Train and Export Inference Graph.

You may have already worked on an image classifier, maybe the Cats and Dog Classification problem, or digit recognition. And this tutorial will take you a step further by covering image classification along with object localization which means you will learn to train a model capable of identifying an object as well as locating it in the image using Keras. 

It’s fine even if you have not worked on a Classification system as it is not a pre-requisite for this tutorial and I have explained everything from scratch in the tutorial. The approach used, is only recommended if you don’t want to, or cannot use the TensorFlow Object Detection (TFOD) API. 

In case you are thinking that video classification would be the same as a classification of a series of images and you can skip this tutorial, then you are totally wrong as in videos the series of frames/images are related to one another which makes video classification a lot more complex and fascinating as well.

The tutorial is about how to perform classification on videos, you will learn about several approaches that can be used to perform video classification, and will implement a simple but quite effective approach called Single-Frame CNN using a moving average technique across frames for human activity recognition. The outline:

• • • Understanding Human Activity Recognition.
    • Video Classification and Human Activity Recognition – Introduction.
    • Video Classification Methods.
    • Types of Video Classification problems.
    • Making a Video Classifier Using Keras. (Moving Average and Single Frame-CNN)

This tutorial will take you a step further by teaching you to implement a much more effective video classification approach for human action recognition using a Convolutional Neural Network combined with a Long-Short Term Memory Network. We’ll be using two different architectures and will also cover some theory on video classification and different approaches that can be used for video classification. The outline:

• • • Image Classification
    • What Exactly Videos are?
    • Video Classification Approaches
    • Human Action Recognition Implementation 
      

Well done! You are doing great, just one more section to go.

This tutorial will serve as a dlib library crash course for you and will cover most of the prominent features and algorithms present in the dlib library, sounds impossible to cover everything in a blog post right?

So I will just provide all the best and the most important tutorials on different aspects of the dlib library out there in a nice hierarchical order as everything is already out there. It will serve as a very well-structured reference guide for dlib library users. The outline:

• • • Dlib Library Installation
    • Face Detection
    • Facial Landmark Detection
    • Facial Landmark Detection Applications (Blink, yawn, smile detection & Snapchat filters)
    • Facial Alignment & Filter Orientation Correction
    • Head Pose Estimation
    • Single & Multi-Object Tracking with Dlib
    • Face Swapping, Averaging & Morphing
    • Face Recognition
    • Face Clustering
    • Training a Custom Landmark Predictor
    • Training a Custom Object Detector
    • Dlib Optimizations For Faster & Better Performance

This tutorial will teach you to train a custom Hand Detector with Dlib, cleverly automate the data collection & annotation step with image processing, and control applications like Games and Video Players with hand gestures, fascinating right? So let’s dive into it. The tutorial can be split into the followings parts:

• • • Part 1: Training a Custom Hand Detector with DLIB.
    • Part 2:  Integrating Gesture controls with Applications.

In this tutorial, you will learn to control the infamous Chrome’s Dino game that appears on Chrome when the internet is not working, using face gestures. 

The game is normally played to kill spare time when the internet is not working, But why not put your time to more productive use, you can go over this tutorial and try to create a similar application whenever your internet stops working. The tutorial is divided into the followings steps:

• • • Step 1: Real-time Face Detection
    • Step 2: Find the landmarks for the detected face
    • Step 3: Build the Jump Control mechanism for the Dino
    • Step 4: Build the Crouch Control mechanism
    • Step 5: Perform Calibration
    • Step 6: Keyboard Automation with PyautoGUI
    • Step 7: Build the Final Application

Congratulations! you have made it to the end and became a proficient practitioner, you must be proud of yourself. But still, I will recommend you to continue your learning journey and keep exploring new technologies, so that you can keep up with others as Computer Vision and Machine Learning are very competitive and rapidly growing fields.

 

It is a free 10-day email course on making a career in computer vision, that will help you reach your desired outcomes, whether it’s landing a job, becoming a researcher, building projects as a hobby, or whatever it might be, this course will help you get that and it’ll show you an ideal path from start to finish to master the computer vision career roadmap.

You can join the course here.

It’s a high-level course that will teach you everything you need to get started with computer vision that is completely free. The course is designed in a way that people of all skill levels can benefit from it. So you don’t need any background in computer vision or Artificial Intelligence for this course. The released episodes are:

• • • Computer Vision For Everyone: Course Introduction (Episode 1 | CVFE)
    • What Is Computer Vision – A very Simple Explanation (Episode 2 | CVFE)
    • Artificial Intelligence: 4 Levels of Explanation Part 1 (Episode 3 | CVFE)
    • History of AI, Rise Of Machine Learning and Deep Learning | Artificial Intelligence Part 2/4 (Episode 4 | CVFE)
    • Different Branches of Machine Learning | Artificial Intelligence Part 3/4 (Episode 5 | CVFE)

And that’s not all, as more episodes are on the way.

It’s a free 25 Page Computer Vision Resource Guide in which you will find some of the best resources (Courses, Books, Tutorials, Blogs, and YouTube Channels, etc) out there, to learn computer vision.

You can download it here.