Covers: theory of convolutions

- Why to use Convolutions?
- What is an image filter?
- What is image segmentation and how is it done?
- What convolution does really do?
- What are two famous and edge detection filters / algorithms?

In this video you will learn about:

- Why to use Convolutions?
- What is image segmentation and how is it done?
- What convolution does really do?
- Edge detection: Sobel Operator
- Edge detection: Scharr Operator

For data with hig dimensionality, for example images (3D but 2D for every RGB channel), but also time-series, a single input for a NN may be respresented by **thousands of data points**. Specifically, a gray image of dimensionality 224×224 contains over 50K pixels! Imagine image classification task. Would each pixel in the image be correlated with the subject/class in the picture? Rather, the networks should itterativelly be trained to associate particular areas of the image capturing at least some part od the subject/class .

Look at the image bellow! Segments or objects, if you want, are masked with different colors. E.g. people - red mask, cars - blue mask? How it is done? By convolution! If you are interested to learn more about segmentation, you may be interested in Mask R-CNN, one of the image segmentation state-of-the-art model.

First, it reduces dimensionality of the input data, so called down-pooling! It uses the kernel operator - a matrix - sliding on the input data. By learning kernel weights it extracts properties in the data/image like edges.

**It's the kernel!** ! The sliding matrix on the matrix (data). In other words, it is a matrix applied to an image in order to recognize patterns.

e.g. you can highlight edges applying a sobel filter through matrix multiplication

Fail to play? Open the link directly: https://youtu.be/7MzPOxL7zlk

Amir Hajian

0 comment

Contributors

- Objectives
- Feature extraction via convolution kernels
- Potential Use Cases
- Building a DL architecture for for computer vision model.
- Who is This For ?
- INTERMEDIATE

Click on each of the following **annotated items** to see details.

Resources6/7

VIDEO 1. Mathematics of Convolution

- Why wee need more than MLP?
- What is a convolution and kernel?
- What is a Hann function and how to apply it?
- What are ConvNets and what is their architecture?

31 minutes

VIDEO 2. Why Convolutions? Sobel & Scharr Filters

- Why to use Convolutions?
- What is an image filter?
- What is image segmentation and how is it done?
- What convolution does really do?
- What are two famous and edge detection filters / algorithms?

16 minutes

VIDEO 3. 2D Convolutions, Pooling, and Dilated Convolutions

- What is pooling and padding?
- What other 2D convolution techniques do exist?
- When do we use padding and how?
- When do we use pooling and how?
- What are dilated convolutions and how they differ from standard convolutions?

30 minutes

VIDEO 4. Conv-Nets

- How can I apply Sobel and Scharr Operator on image?
- What is the difference between Sobel and Scharr Operator and how can I visually compare them?
- What is a Convolutional Neural Network (CNN)?
- How can one understand a CNN?

20 minutes

VIDEO 5. Regularization using Dropouts

- Why do we need dropouts?

17 minutes

REPO 6. Hands-on Convolutional Networks

- How to implement convnets in PyTorch?

30 minutes

RECIPE 7. Understanding Convolution

40 minutes

0 comment