Video 67: Local Binary Pattern (LBP)

 

Hello viewers. In this video, a very popular texture descriptor Local Binary Pattern (LBP) is explained. Here its basic theory, calculation method, its performance analysis and MATLAB implementation is given.  This video includes following components,

• Introduction to Local Binary Pattern (LBP).

• How an LBP feature is obtained from an image.

• Implementation of LBP in MATLAB (with MATLAB Code).

• Comparative performance analysis of LBP.

• LBP as feature vector for Machine Learning applications.

Download Timo Ojala Paper: Click here

Read More

Video 65: Texture Classification using Wavelet Scattering Transform (with MATLAB Code)

 

Hello viewers. In this video, Texture Classification is presented based on Wavelet Scattering Transform (WST). WST is also briefly explained. This lecture explains that how WST coefficients can be used as feature vectors for classification task. Texture images are taken from KTH_TIPS and KYBERGE image databases. 

This video includes following components,

* Brief introduction to Wavelet Scattering Transform (WST).

* Computing Image Features (WST Coefficients as Features).

* Texture Image Databases used.

* Training Algorithm.

* Testing Procedure.

* MATLAB Implementation (with MATLAB Code).

Download:

Texture Image Database: Download

Link of previous video:

1. Introduction to Wavelet Theory and Its Applications: Click Here

2. Wavelet Scattering Transform for Signals and Images: Click Here

Read More

Video 62: Color Edge Features and DWT based Image Retrieval (With MATLAB Code)

 

Hello viewers, in this video, Content Based Image Retrieval (CBIR) is implemented. This CBIR utilizes both the color and edge features of the images. For this purpose, Color Edge Histograms are obtained. To reduce the size of feature vector, Discrete Wavelet Transform (DWT) is also used. The simulation results show the effectiveness of the proposed algorithm for effective CBIR.     

 

This video includes following contents, 

* Introduction to Content Based Image Retrieval (CBIR).

* Color Edge Feature (Proposed  Algorithm).

* Finding Feature Vector (Training Process).

* Testing Process.

* MATLAB implementation (with MATLAB code).

* Result Analysis.

-----------------------------------------------------------

1. Previous video:

   Color Layout Descriptor (CLD) of MPEG7 for Image Retrieval: Click Here

   

2. Previous video:

   Edge Histogram Descriptor (EHD) of MPEG7 for Image Retrieval: Click Here

   

3. Previous video:

   Content Based Image Retrieval (CBIR) using Wavelet features, CLD and EHD of MPEG7: Click Here

---------------------------------------------------------------

Download Resources:

Image Database: Click Here

Read More

Video 61: Time Series Prediction using ANFIS (With MATLAB Code)

 

Hello viewers, in this video, The Time Series Prediction using Adaptive Neuro-Fuzzy Inference System (ANFIS) is explained. The time series taken here is Mackey-Glass chaotic time series, which is considered as benchmark problem. The ANFIS based algorithm for time series prediction is explained in detail.     

 

This video includes following contents, 

* Introduction to time-series prediction.

* ANFIS for time-series prediction.

* Mackey-Glass chaotic time series (A benchmark).

* Time series prediction algorithm using ANFIS.

* MATLAB implementation (with MATLAB code).

* Result Analysis.

-----------------------------------------------------------

1. Previous video:

"Fuzzy Logic Controller (FLC)": Click Here

2. Previous video:

"ANFIS (Adaptive Neuro Fuzzy Inference System)": Click Here

 

Read More

Video 60: ANFIS: Neuro-Fuzzy Inference System (Theory and MATLAB Implementation)

 

Hello viewers, in this video, The Neuro-Fuzzy modelling highlighting ANFIS is explained. The basic theory of ANFIS is presented. Also the complete process of MATLAB implementation is given. In MATLAB implementation, the ANFIS is used as universal approximator. The two functions 1D sin(t) and 2D sin(r)/r are realized using ANFIS.   

 This video includes following contents, 

• Neuro – Fuzzy Modelling.
• Adaptive Neuro-Fuzzy Inference System (ANFIS).
• ANFIS Architecture.
• ANFIS Hybrid learning algorithm.
• ANFIS Applications.
• ANFIS as Universal Approximator (UA).
• MATLAB Implementation of ANFIS as UA (with MATLAB code).

--------------------------------------------------------------------------------------

1. Previous video:

"Fuzzy Logic Controller (FLC)": Click Here

2. Link for research paper of Jang: Click Here 

Read More

Video 57: Rice Grain Quality Assessment using Morphological Image Processing (Offline and Realtime mode)

 

Hello viewers, in this video, an implementation of rice quality assessment algorithm is shown. The proposed algorithm is based on morphological image processing. The quality of rice is estimated only on the basis of the length of rice grains. This implementation is done for both offline and online (Realtime) mode. In offline mode, images of rice grains are given to the algorithm and in real time mode, a camera is attached which takes images of rice grains and tells the outcome in real time.

 

This video includes following contents, 

• Need of automated rice grain quality assessment. 
• Steps of proposed scheme based on morphological image processing.
• Solving touching grains problem.
• MATLAB Code for proposed scheme.
• Testing in offline and online mode.

Download

Test Images: Download

Read More

Video 54: Face Recognition using Wavelet Features and PCA (With MATLAB Code)

Hello viewers, In this video, a face recognition scheme is implemented using Wavelet Features and Principal Component Analysis (PCA). Here wavelets are used to extract facial features and PCA is used to reduce the size of wavelet feature vectors. The proposed scheme is very robust and capable to recognize the faces even some changes occur in faces such growing beard and mustache or putting goggles etc.

This video covers followings contents,

• Face image database (Faces94).
• Face image database preparation for training and testing.
• Finding wavelet features.
• PCA for dimension reduction.
• Training and testing procedures.
• MATLAB Code for Training.
• MATLAB Code for Testing (Discrete and Bulk).

-------------------------------------------------------------------------------------------------------------------------

Links of previous videos:

1. Principal Component Analysis (PCA) for Images and Signals: Click Here

2. Face Recognition using PCA in MATLAB: Click Here

Links for Face Image Database:

1. https://cmp.felk.cvut.cz/~spacelib/faces/ (All Faces)

2. https://cmp.felk.cvut.cz/~spacelib/faces/faces94.html (Face94 database)

3. https://www.exptech.co.in/p/resource-download.html (Database used in video)

Download Resources:

1. All Image Database: Download

2. Test Images for Robustness: Download

Read More

Video 53: ECG based Heart Disease Diagnosis using Wavelet Features and Deep CNN (Arrhythmia detection)

Hello viewers. This is recorded video of an invited guest lecture delivered by me in an international conference held in July 2021.

This video is about heart disease diagnosis using wavelet features and deep CNN mainly focusing on Arrhythmia detection and heart rate estimation.

This video includes following contents.

• Introduction (Problem Statement).
• Basics of ECG signals and QRS Complex. 
• ECG Database on PhysioNet.
• Proposed wavelet based algorithm for Heart Rate estimation and Arrhythmia detection.
• Deep CNN based approach of Arrhythmia detection.
• Conclusions.

This lecture is based on my previously published YouTube videos, which you can find on following links.

1. ECG signals Classification using CWT and Deep Neural Network in MATLAB: Click Here
2. ECG's QRS Peak Detection and Heart Rate Estimation using Discrete Wavelet Transform (DWT) in MATLAB: Click Here

Other Links:

1. PhysioNet: https://physionet.org/

2. PhysioNet Databases: https://physionet.org/about/database/

3. PhysioNet Bank ATM: https://archive.physionet.org/cgi-bin/atm/ATM

4. ECG signal database GitHub repository:

     https://github.com/mathworks/physionet_ECG_data/

Read More

Video 46: Object Classification using HOG features and ECOC Multi-Class SVM (With Matlab Code)

Hello Viewers, in this video, a multi-class object classification problem using HOG features is explained. To demonstrate the implementation, simple geometrical shapes (Circle, Square, Star and Triangle) are taken for classification. As a classifier, ECOC (Error Correcting Output Codes) based multi-class SVM is used. The shapes image database is obtained from Kaggle. 

The HOG feature is very popular and widely used for object detection in images. To understand the HOG feature computation, viewers are requested to watch my previous video of HOG feature computation.

This video includes following contents:

• Introduction.
• Proposed scheme for object Classification.
• Image Database Preparation.
• ECOC based Multi-Class SVM.
• Appropriate Cell Size selection for HOG feature.
• MATLAB Code for Shapes Classification (Multi-Class).
• MATLAB Code for Discrete Testing.

1. Link for previous video on HOG feature computation: Click Here

2. Link for Kaggle Dataset: https://www.kaggle.com/smeschke/four-shapes

3. Link to download original paper of N. Dalal and Bill Triggs:

https://lear.inrialpes.fr/people/triggs/pubs/Dalal-cvpr05.pdf

Download Resources: 

1. Image Dataset (Modified): Download (Source: Kaggle)

2. Distorted Test Images: Download

Read More

Video 41: Detecting STOP Traffic Sign using Deep RCNN (Real time and Offline mode)

Hello Viewers, in this video, It is explained that how one can implement a deep RCNN for detecting 'STOP' traffic signs from image and videos in both offline and real time mode.

Also, it is shown that how one can use MATLAB's Labeler app to create database for training.

Here, transfer learning is used and for fine tuning, a pre trained RCNN is re trained on our own image dataset which is created with help of Image Labeler app.

This video has following contents:

• Introduction.
• What is Deep R-CNN?
• Proposed Scheme using Deep R-CNN.
• Understanding MATLAB’s Image Labeler for image database creation.
• MATLAB Code for training.
• MATLAB Code for testing.
• Code execution and result analysis.

Links of previous videos:

1. How to create a Deep CNN: Click Here

2. ECG signals classification using wavelet features and deep CNN: Click Here​

3. Implementing deep CNN in Python using TF and Keras (Face Mask detection): Click Here

Download Resources: 

1. Test Video: Download (Source: Mathworks.com)

2. Traffic Sign Image Database: Download

Read More

Video 39: Implementing Deep CNN in Python using TensorFlow and Keras (Face Mask Detection Problem)

Hello Viewers, in this video, it is shown that how a deep Convolutional Neural Network (CNN) can be implemented in Python using TensorFlow (TF) and Keras (K).

To understand its working, an interesting example is taken for implementation, where we can classify face images 'With Mask' and 'Without Mask'.

This video has following contents:

• Objective: Recognize a face with Mask or without Mask.
• Proposed Deep CNN based scheme.
• Introduction  to TensorFlow, Keras and Python.
• Image Dataset used.
• Python Code to implement deep CNN using TF and K.
• Simulation result analysis.

Links of previous videos:

1. How to create a Deep Neural Network in MATLAB: Click Here

2. Introduction to Deep Learning: Click Here

3. Anaconda Download: https://www.anaconda.com/products/individual#Downloads

4. Image Dataset: https://www.kaggle.com/ashishjangra27/face-mask-12k-images-dataset

Read More

Video 24: ECG Signals Classification using Continuous Wavelet Transform (CWT) & Deep Neural Network in MATLAB

Hello Viewers, in this video, ECG signals are classified using pretrained deep CNN such as AlexNet via transfer learning. As we know that AlextNet can accepts input as image only, therefore, it is not possible to give 1D ECG signals to AlexNet directly. 

To solve this problem, we utilize the strength of Continuous Wavelet Transform (CWT) to represent 1D ECG signals into image, so that it can be fed as input to deep CNN AlexNet. 

Using CWT, we obtain CWT coefficients of 1D ECG signal and these coefficients are arranged as scalogram to represent in form of image. The ECG database is taken from Physionet.

This video has following contents:

• Types of ECG Signals for Classification.
• ECG Signal Database.
• Converting 1D ECG signals to Image using CWT Scalogram.
• Transfer Learning via pretrained AlexNet deep CNN.
• MATLAB Code for CWT Scalogram Image database creation.
• MATLAB Code for AlexNet Training and Validation.

Important Links:

1. Continuous Wavelet Transform of 1D signals using Python and MATLAB: Click Here

2. How to create a deep neural network in MATLAB : Click Here

3. ECG signal database GitHub repository: https://github.com/mathworks/physionet_ECG_data/

Read More

Video 22: How to Create a Deep Neural Network in MATLAB (Digit Recognition Example)

Hello viewers, In this video, It is explained that how one can create a deep neural network such as Convolutional Neural Network (CNN) in MATLAB. All the layers are also explained in details with their structure. This CNN will be trained with images of handwritten digits of MATLAB's dataset. 

• MATLAB’s Digit Dataset.
• Digit Dataset Preparation.
• Structure of a Convolutional Neural Network (CNN).
• Explanation of layers of a CNN and Training Parameters.
• MATLAB Code of Training and Validation.
• MATLAB Code of Discrete Testing.

Important Links:

1. Introduction to Deep Learning: Click Here

2. MNIST Dataset (Digit and Fashion) to PNG/JPG Images Conversion using MATLAB: Click Here

Download Resources: 

Digit Database: Download

Read More

Video 21: MNIST Dataset (Digit & Fashion) to PNG/ JPG Images Conversion using MATLAB

Hello viewers, In this video, It is explained that how MNIST dataset which is in complex format (idx-ubytes and csv) can be converted in to simple png/ jpg images in structured folders. So that it becomes easy to visualize the dataset and to have an idea that what types of images we actually have in dataset for training and testing. For this purpose, two popular MNIST datasets are considered for conversion, 

1. Handwritten Digit Dataset and

2. Fashion Dataset.

The datasets are downloaded in CSV format from Kaggle for conversion.

This video covers followings contents,

• MNIST Digit dataset (Yann Le and Kaggle link).
• MNIST Fashion dataset (GitHub link and Kaggle Link).
• Need of dataset as PNG/ JPG images.
• Conversion logic from csv to PNG/ JPG images and folder preparation.
• MATLAB Code for conversion of Digit dataset to PNG/ JPG images.
• MATLAB Code to conversion of Fashion dataset to PNG/ JPG images.

Important links:

1. Digit Dataset,

    Yann Le page (idx-ubytes): http://yann.lecun.com/exdb/mnist/​

    Kaggle Page (csv): https://www.kaggle.com/oddrationale/mnist-in-csv

2. Fashion Dataset,

    GitHub Page (idx-ubytes): https://github.com/zalandoresearch/fashion-mnist

    Kaggle Page (csv): https://www.kaggle.com/zalando-research/fashionmnist

Read More

Video 20: Introduction to Deep Learning

Hello viewers, In this video, A brief introduction of Deep Learning (DL) is presented. Also Supervised, Unsupervised and Reinforcement learning are also discussed. Convolutional Neural Networks are also explained and various popular deep neural architectures such as Alexnet, Googlenet, Resnet, RCNN and YOLO are presented. In the last, training a deep neural model is also explained via transfer learning and training from scratch. 

This video covers followings contents,

• Branches of AI and Branches of ML.
• Supervised, Unsupervised and Reinforcement Learning.
• Deep Learning.
• Deep Learning Vs. Machine Learning.
• Convolutional Neural Network (CNN).
• Deep Learning Architecture (AlexNet, GoogleNet, VGG Net, ResNet, RCNN, YOLO) 
• How to create Deep Learning Model?
• Training from scratch and Transfer Learning.

Read More

Video 19: Introduction to Artificial Intelligence (AI around us)

Hello viewers, In this video, A brief introduction of Artificial Intelligence (AI) is described emphasizing its applications. AI is now impacting our day to day life. We use AI in many forms in our daily life whether we are aware of it or not. 

This video covers followings contents,

• What is AI?
• Levels of AI.
• AI in Smart Phones.
• AI in Smart Cars.
• AI in UAVs (Drones).
• AI in Social Media.
• AI in Media Streaming Services.
• AI in Video Games.
• AI in online Ad networks.
• AI in Navigation and Travel.
• AI in Banking and Finance.
• AI in Smart Homes.
• AI in Security and Surveillance.
• AI in Medical.
• AI in Other Areas.
• Branches of AI

Read More