A short quiz on humans (mostly faces)

Adrian F. Clark

Which of the following techniques is "eigenfaces" built on?
- Linear Discriminant Analysis
- Convolutional Neural Network
- Principal Component Analysis
- Support Vector Machine
Eigenfaces is based on a Principal Component Analysis or Eigen decomposition of the training data. The other methods listed are also machine learning techniques and some of them may be used in conjunction with PCA to recognise faces, but the resulting technique is not eigenfaces.
What is the basis of the HOG feature detector?
- a histogram of edge gradients
- a histogram of edge orientations
- a histogram of pixel values
- a histogram of colour differences
HOG stands for "histogram of oriented gradients", so the histogram is of edge orientations, also known as edge directions.
What is covariance?
- a pair of images having similar variances
- a measure of the similarity of variations in the images
- another name for cross-correlation
- a pair of images having dissimilar variances
The covariance measures whether variations in images happen "in step". For two images $P$ and $Q$ with means $\bar P$ and $\bar Q$, it is $\sum (P_{ij} - \bar P) (Q_{ij} - \bar Q)$.
Faces are allegedly more attractive if:
- the lips are oval in shape
- the shape matches the golden ratio
- facial features are slightly asymmetrical
- the hair and skin are in the same region in HSV space
People are supposed to have more beautiful faces if the shape is closer to the golden ratio. No, I don't believe it either.
What does Principal Component Analysis (PCA), as used in eigenfaces, do?
- splits up features into components
- finds the directions in feature space with the largest variations
- finds the most similar set of features in a database
- finds the most important direction in an image
PCA finds the direction in feature space with the largest variance, then the next-largest direction that is perpendicular to the first -- and so on. For the mathematically-inclined, this forms a set of basis vectors to the feature space using the criterion of maximising variance.
What is Affective Computing?
- taking into account the emotional state of the user
- a way of recognising facial expressions
- a way of performing computer vision that is more accurate
- a face recognition algorithm
Affective Computing attempts to take account of the emotional state of the user. Much of the research into affective computing directed towards identifying the expression on a user's face, but there are other techniques for doing this (skin conductivity, ECG, etc) and the subject also covers things that can be done when this state is known.
What are Haar features?
- measures of the number of edges in a region
- the sum of differences between a pair of images
- the number of corners in a region
- differences between the sums of pixels in rectangular regions
Haar features are simple, which makes them quick to calculate.
What is Adaptive Boosting?
- a way of improving the contrast of an image based on the local grey levels
- a method for improving the performance of a single weak classifier
- a method for combining multiple weak classifiers into one strong classifier
- an adaptive way of choosing the best single classifier for a task
Adaptive booting is the way that the Viola-Jones algorithm combines multiple Haar features, each of which is a weak classifier individually, into a strong classifier
Why is the integral image representation used in Viola-Jones?
- it allows Haar features to be calculated in constant time
- it stores where faces lie in an image
- it stores the sum of all the pixels in an image
- it allows features to be concatenated into a feature vector
The main attraction the integral image representation is that it allows Haar features of any size to be calculated in constant time, using the same underlying mathematics as the anti-aliased texture mapping on a typical graphics card.
How is Principal Component Analysis performed?
- by Eigen decomposition of the covariance matrix
- by Eigen decomposition of the variance of an image
- by Linear Discriminant Analysis of the variance of an image
- by Linear Discriminant Analysis of the covariance matrix
PCA is performed by calculating the covatiance matrix and then performing an eigen decomposition of it. The resulting eigenvectors or Principal Components describe the different ways in which the images vary and can be used for "transforming" an image. When that is done, the eigenvalues are a set of weights that describe how the principal components are combined to give that image. These weights can be used for recognition: the closer they are to something in the training image set, the more similar they are.