Feature engineering

 Duration: 4 hrs
 Outline:
1. Introduction
2. Feature engineering
3. Features in visual pattern recognition
4. Shape-based feature descriptors
 Feature engineering

 Duration: 4 hrs
 Outline:
1. Introduction
2. Feature engineering
3. Features in visual pattern recognition
4. Shape-based feature descriptors
Introduction to feature & feature engineering
 Feature:
 an individual measurable property or characteristic of a data
example

 describes the example

 Features are usually numeric.

 Feature engineering: transfer raw data into feature vector

Data  feature vector  ML model

The general framework for Machine Learning
 Curse of dimensionality
 Dimensionality: the number of features in feature vector.

 Curse of dimensionality:

 The number of features is very large relative to the number of

observations (examples) in dataset

 Hard to train effective model

 Dimensionality reduction
 Feature selection
 Feature extraction
 Feature extraction vs. feature selection

 Feature selection:

 Filtering irrelevant or redundant features from dataset

 Choosing a subset of the original features

 Feature extraction:

 Creating a new smaller set of features

 Getting useful features from existing data

 Feature need to be informative, discriminating and

independent
Feature extraction vs. feature selection
 Feature engineering

 Duration: 4 hrs
 Outline:
1. Introduction
2. Feature engineering
3. Features in visual pattern recognition
4. Shape-based feature descriptors
 Feature engineering
 One-hot encoding

 Binning

 Normalization

 Standardization

 Dealing with missing feature

 Data imputation techniques

 One-hot encoding
 Transform a categorical feature into several binary features

 Example: feature “color” has 3 values “red”, “yellow”, green”

 “red” = 1, “yellow” = 2, “green” = 3

 ?
 Binning (bucketing)
 Transform a numerical feature into categorical feature

 Example: feature “age”

 Put all ages between 0 and 5 years-old into one bin

 Put ages from 6 to 10 years-old in the second bin

 Put ages from 11 to 15 years-old in the third bin, and so on.

 Normalization

 Converting an actual range of values of a numerical feature

into a standard range of values, typically in the interval [-1, 1] or
[0, 1].

 Example: natural range = [350, 1450]

 Subtracting 350 from every value of the feature

 Dividing the result by 1100  normalized range = [0, 1].

 Standardization

 Rescaling the feature values so that they have the properties of

a standard normal distribution with µ = 0 and =1

 Formula:
 Standardization or normalization?
 Try two if have time 

 Rule of thumbs:
 Dealing with missing features

 Removing the examples with missing features.

 Use data imputation technique

 Data imputation techniques

 Technique 1: Replacing the missing value of a feature by an

average value of this feature in the dataset

 Technique 2: Replacing the missing value by the same value

outside the normal range of values.

 Technique 3: Replacing the missing value by a value in the

middle of the range.

…etc…
 Feature engineering

 Duration: 4 hrs
 Outline:
1. Introduction
2. Feature engineering
3. Features in visual pattern recognition
4. Shape-based feature descriptors
 Image feature extraction
 Purpose:

 To reduce the dimensionality of input image

 To transform each input image into a corresponding multi-
dimension feature vector

 To perform the predefined classification tasks with sufficient

accuracy without using the entire input image

 Requirements:
 Features should extract the most suitable characteristics from
the input image
An example of feature extraction

.
 Visual features

 Color-based features
 Visual features
 Shape-based features
 Visual features
 Texture-based features
 Which feature is the best?

 Example: plant recognition

 Plant features: leaf, fruit, flower, root, branch,…

 Leaf features: shape, vein, margin, texture

 No single best feature for a given leaf identity  combination
of different features

 No single best presentation for a given feature  multiple

descriptors to characterize the feature from different
perspectives

 Challenging
Deep learning
 Innovative
 Feature engineering

 Duration: 4 hrs
 Outline:
1. Introduction
2. Feature engineering
3. Features in visual pattern recognition
4. Shape-based feature descriptors
 Shape-based feature descriptor

 Shape: important

 Good shape descriptor: invariant to geometrical

transformations (rotation, reflection, scaling, translation)

 Types of shape descriptors: simple and morphological shape

descriptor (SMSD), contour-based, region-based
Simple and morphological shape descriptor

 Refer to basic geometric properties of the shape

 Basic descriptor: diameter, major axis length, minor axis length,

area, perimeter, centroid,…

 Morphological descriptor: aspect ratio, perimeter to area ratio,

rectangularity measures, circularity measures,…
 Contour-based feature descriptor

 Consider the boundary of a shape and neglect the information

contained in the shape interior

 Ex: CCD (centroid contour distance), Fourier descriptor

computed on CCD.
Contour-based feature descriptor
 Region-based feature descriptor

 Take all the pixels within a shape region into account to

obtain the shape representation

 Image moments: statistical descriptor of a shape. Ex: Hu

moments

 Local features: select key points in image. Ex: HOG

(histogram of oriented gradients), SIFT (scale-invariant
feature transform)
 Histogram of Oriented Gradients (HOG) ALGORITHM

• HOG stands for histogram of oriented gradients.

• The hog descriptor focuses on structure or shape of
the object.
• It uses magnitude as well as direction of the
gradient to compute the features.
• It generates histogram by using magnitude and
direction of the gradient.
HOG ALGORITHM
 HOG ALGORITHM

20

40 70

• Here we calculating gradient magnitude and 70

direction, to calculate pixels intensity we need
• X direction=|40-70|=30
• Y direction=|20-70|=50
• By these values we are calculating magnitude and
direction of the gradient
• By using magnitude and direction we calculate
feature vectors
 HOG ALGORITHM
e
 HOG ALGORITHM

• Before getting the hog feature and after

concatenating feature vectors we are supposed to
do normalize.
• Suppose we have taken 150*300 pixels and multiply
with 2 to increase the brightness and divided by 2 to
decrease the brightness, then you cant compare
two images without normalization bec’z the pixels
intensity will be changed.
• But if you normalize the feature vectors it is easy to
compare
HOG ALGORITHM
 HOG ALGORITHM

• For hog features giving human template and giving output

for convolving with human model
• Then it will predict whether it is human or not.
Image moments
 Hu moments feature descriptor
Central moments:

Central normalized moments:

Centroid of the image:

_
x.s(x, y) _
 y.s(x, y)
x x y
,y x y

s(x, y) s(x, y)
x y x y
Hu moments feature (cont)
Id image S1 S2 S3 S4 S5 S6 S7

6
 HOG

https://www.youtube.com/watch?v=XmO0CSsKg88&t=41s