Introduction to the course5:40
Presenting what you are going to learn in this course; and what you will be able to do by the end of the course. Illustrating that this course will be fun and engaging with a lot of coding activities.
Quick Win #1: Convolution34:09
Implement convolution operation to grayscale image. Detect edge of the object on image by different filters.
Coding Activity: Convolution
Quick Win #2: Pooling16:11
Apply max pooling operation to grayscale image. Demonstrate downsampled output image.
Coding Activity: Pooling
Quick Win #3: Convolution+Pooling23:51
Combine two operations together. Plot resulted output images.
Coding Activity: Convolution + Pooling
Quick Win #4: Convolution in Real Time by camera24:35
Demonstrate convolution in Real Time by camera. Visualize object contour. Compute and draw bounding box by contours coordinates.
Coding Activity: Define a 3x3 filter
Quick Win #5: Track movement of the object via Convolution13:28
Track object via convolution in Real Time. Calculate object centre and visualize tracker line.
Coding Activity: Update deque object
Glossary3:15
Generate glossary with key terminology. Fill it out continuously throughout the course.
Software Installation & Verification47:28
Install needed prerequisites for the course. Verify successful installation.
How to study the course?2:41
Recognize the recommended way to study this course. Investigate tips to obtain best possible experience.

Introduction & Learning objectives: Assembling dataset1:52
How to create own dataset for classification tasks? Identifying objectives for the Section.
Toolkit to download images8:48
Set up toolkit to get images from large and existing dataset. Verify successful installation.
Download images from large and existing dataset by toolkit14:31
Apply toolkit to obtain needed classes of objects. Differentiate commands to download dataset with specific parameters.
Activity: Download images for a given class1:39
Implement command to download limited number of images. Specify attributes to obtain dataset with given parameters.
Modify downloaded dataset to use it for Classification30:39
Cut objects from images to use them for classification. Assemble and save prepared dataset.
Coding Activity: Pandas dataFrame
Download other datasets5:17
Obtain dataset of handwritten digits MNIST and dataset with colour images of 10 classes CIFAR.
Process other datasets to use them for Classification29:29
Modify images of MNIST and CIFAR to use them for classification. Assemble and save prepared datasets.
Coding Activity: Splitting Dataset
Quiz: What are the best practices for creating own dataset?
Conclusion: key takeaways for assembling custom dataset2:33
Recap and summarize the Section. Sense the progress and transit to the next step.

Introduction & Learning objectives: Modifying existing dataset1:26
How to convert dataset into needed format for classification tasks? Identifying objectives for the Section.
Download dataset of Traffic Signs5:17
Obtain dataset GTSRB of Traffic Signs. Investigate its original structure.
Convert downloaded dataset to use it for Classification25:07
Modify images of Traffic Signs to use them for classification. Save prepared dataset.
Coding Activity: Pandas dataFrame
Quiz: What are the best practices for converting dataset?
Conclusion: key takeaways for modifying existing dataset2:03
Recap and summarize the Section. Sense the progress and transit to the next step.

Introduction & Learning objectives: Applying preprocessing approaches1:15
How to implement different preprocessing approaches on prepared datasets before training? Identifying objectives for the Section.
Construct set of datasets with colour images18:00
Produce datasets from prepared ones by applying variety of preprocessing techniques. Save set of processed datasets in colour.
Coding Activity: Mean Image
Construct set of datasets with grayscale images23:04
Produce datasets from prepared ones by applying variety of preprocessing techniques. Save set of processed datasets in grayscale.
Coding Activity: Standard Deviation
Quiz: What are the best practices for implementing preprocessing approaches?
Conclusion: key takeaways for applying preprocessing techniques2:47
Recap and summarize the Section. Sense the progress and transit to the next step.

Introduction & Learning objectives: Designing deep architectures1:32
How to construct deep architectures for CNN models? Identifying objectives for the Section.
How many Convolutional-Pooling pairs of layers?55:07
Select deepness of network by number of convolutional and pooling layers in a sequence. Interpret notation.
How many Feature Maps in Convolutional layers?33:11
Examine number of feature maps for every convolutional layer. Interpret notation.
How many Neurons in Fully Connected layer?32:01
Research needed amount of neurons for output layer. Interpret notation.
How much Dropout?34:48
Analyse percentage of dropout after every layer. Interpret notation.
What else?41:12
Improve designed deep network by adding advanced features. Interpret notation.
Save designed deep CNN models into binary files19:57
Save all designed deep networks into binary files. Visualize and save structure of the models.
Quiz: What are the best practices for constructing deep CNN architectures?
Conclusion: key takeaways for designing deep CNNs2:32
Recap and summarize the Section. Sense the progress and transit to the next step.
Heuristic approach to identify the best model7:15
How to train every model 100 times at every stage?

Introduction & Learning objectives: Training developed deep CNNs models1:44
How to train, test and differentiate the best deep CNN models? Identifying objectives for the Section.
Overfit designed deep models with prepared datasets24:04
Implement overfitting with small amount of images from prepared datasets. Plot resulted charts.
Train designed deep models with prepared datasets1:23:56
Run training process for all developed models and all prepared datasets. Save trained weights.
Test trained models1:49:51
Evaluate accuracy of every deep model on testing dataset. Display confusion matrix.
Test Classification in Real Time by camera26:47
Classify new images by camera in Real Time. Plot bar chart with scores in Real Time.
Combine: Detection & Classification in Real Time by camera23:40
Apply simple object detection by colour thresholding in Real Time. Classify detected fragment in Real Time.
Visualize training process of filters28:46
Compose video to show process of training filters for convolutional layer. Demonstrate results.
Quiz: What are the best practices for training and testing deep CNNs models?
Conclusion: key takeaways for training designed CNNs3:22
Recap and summarize the Section. Sense the progress and transit to the next step.

Introduction & Learning objectives: Generating additional artificial data1:28
How to produce additional artificial data for prepared datasets? Identifying objectives for the Section.
Change brightness of images in dataset11:31
Generate additional images by random brightness. Plot resulted images.
Manipulate images by geometric transformations16:21
Generate additional images by rotation and projection. Plot resulted images.
Augment and equalize images in dataset21:33
Add transformed images into dataset. Make number of images equal for all classes. Plot equalized histogram.
Visualize unique examples from augmented dataset13:01
Plot images from augmented dataset. Compose video from plotted examples.
Quiz: What are the best practices for producing additional artificial data?
Conclusion: key takeaways for generating synthetic data1:17
Recap and summarize the Section. Sense the progress.

What does Confusion Matrix show?27:20
Explained theory on what Confusion Matrix shows.
Coding Activity: Confusion Matrix
Quiz: Confusion Matrix
2D Image Convolution: Numpy, Tensorflow, Keras0:34
Explained theory on how 2D Convolutional Layer creates feature maps. Instructions on how to use Tensorflow and Keras functions ‘get_weights()’ and ‘set_weights()’.
Coding Activity: Reshape a given input

Requirements

Basic knowledge of Image Classification Algorithms
Basics on how CNN works
Intermediate knowledge of Python V3
Basic knowledge of OpenCV
Basic knowledge of Tensorflow
Basics on how to use Anaconda Environments
Basics on how to code in Jupyter Notebook

Description

In this practical course, you'll design, train and test your own Convolutional Neural Network (CNN) for the tasks of Image Classification.

By the end of the course, you'll be able to build your own applications for Image Classification.

At the beginning, you'll implement convolution, pooling and combination of these two operations to grayscale images by the help of different filters, pure Numpy library and 'for' loops. We will also implement convolution in Real Time by camera to detect objects edges and to track objects movement.
After that, you'll assemble images together, compose custom dataset for classification tasks and save created dataset into a binary file.
Next, you'll convert existing dataset of Traffic Signs into needed format for classification tasks and save it into a binary file.
Then, you'll apply preprocessing techniques before training, produce and save processed datasets into separate binary files.
At the next step, you'll construct CNN models for classification tasks, select needed number of layers for accurate classification and adjust other parameters.
When the models are designed and datasets are ready, you'll train constructed CNNs, test trained models on completely new images, classify images in Real Time by camera and visualize training process of filters from randomly initialized to finally trained.
At the final step, you'll pass Practice Test according to the all learned material during the course.
As a bonus part, you'll generate up to 1 million additional images and extend prepared dataset by new images via image rotation, image projection and brightness changing.

The main goal of the course is to develop and improve your hard skills in order to apply them for real problems of Image Classification based on Convolutional Neural Networks.

Every lecture of the course has SMART objectives. It means, that you can track your progress and witness practical results within the visible time frame, right after the end of the lecture.

S - specific (the lecture has specific objectives)
M - measurable (results are reasonable and can be quantified)
A - attainable (the lecture has clear steps to achieve the objectives)
R - result-oriented (results can be obtained by the end of the lecture)
T - time-oriented (results can be obtained within the visible time frame)

Who this course is for:

Students who want to build complete application for Image Classification with CNN
Students who want to improve their hard skills on Image Classification with CNN before their next interview for internship or dream job
Students who want to use CNN with their Own Data for Image Classification but don't know where to start
Young Researchers who study different Image Classification Algorithms and want to Train CNN with Custom Data and Compare results with other approaches
Students who know basics of Image Classification but want to know how to Train CNN with New Data
Students who study Computer Vision and want to know how to use CNN for Image Classification
Students who work on project of safety driven and want to Classify Traffic Signs with CNN
Students who develop alarm-warning system for driver and need to Classify Traffic Signs

Convolutional Neural Networks for Image Classification

What you'll learn

Explore related topics

Coding Exercises

Course content

Welcome9 lectures • 2hr 51min

Assemble custom dataset for Image Classification8 lectures • 1hr 35min

Modify existing dataset of Traffic Signs for Classification4 lectures • 34min

Apply preprocessing techniques for datasets before training4 lectures • 45min

Design deep CNNs architectures for efficient Classification9 lectures • 3hr 48min

Train and Test designed CNNs models8 lectures • 5hr 2min

Practice Test2 lectures • 40min

Generate synthetic data to augment datasets6 lectures • 1hr 5min

How does it work?2 lectures • 28min

How to move from image recognition to object detection?1 lecture • 1min

Requirements

Description

Who this course is for: