Face detection

FACE DETECTION
Pritam Banerjee
The University of Texas at Tyler
Department of Computer Science
Tyler, TX, 75799
pbanerjee@patriots.uttyler.edu

Face Detection
 What is face detection ?
 Why is face detection important ?
 Classification of face detection
 Image based face detection
 Knowledge based face detection
 Comparison between Image based & Knowledge
based methods
 Open Issues

What is Face Detection ?
Face detection is to identify and detect all image
regions in an image which contain a face regardless
of its three-dimensional position, orientation, and
lighting conditions, and if there is a face present in
the image, return its location and extent of the face.
The process is complex and challenging because it
needs to detect faces which are non-rigid and have a
high degree of variability in size, shape, color, and
texture.

Face Detection
 Why is face detection important
based methods
 Open Issues

Why is Face Detection Important ?
1) Face detection algorithms are used in a wide
range of applications, such as security control, video
retrieving, biometric signal processing which is often
mixed with face recognition, human computer
interface, face recognition and image database
management.
2) Modern digital cameras automatically focus on the
faces when those are detected. Some of the
cameras can also detect facial expressions such as
a smile and then automatically captures the
photograph without the actual need of a

Why is Face Detection Important ?
3) Face detection is getting immensely
important for marketers as well. A webcam can
be integrated into a television which can detect
the face of any person who walks by. Then the
system can calculate the race, gender, and an
approximate age of the person by analyzing the
person’s facial features. Once the information is
collected, a series of advertisements can be
played that is specific toward the detected race

Face Detection
 Classification of face detection techniques
based methods
 Open Issues

Classification of Face Detection Techniques
Image based techniques
1. Eigenface 1. Template Matching
2. Distribution based 2. Skin Color based detections
3. Support Vector Machines 3. Facial feature based detections
4. Bayes Decision Rule
Knowledge based techniques
Face detection techniques

Image based techniques
Image-Based methods use training/learning methods
to make comparisons between face and non-face
images. For these methods, a large number of images
of faces and non-faces should be trained to increase
the accuracy of the system. Eigenface, Neural
Networks and Support Vector Machines are kinds of
methods that are used commonly in image based face
detection algorithms.

Face Detection
 Knowledge based Face Detection
based methods
 Open Issues

Knowledge based techniques
Knowledge based techniques use information about
facial features, skin color or template matching.
Facial features include eyes, mouth, nose or other
facial features to detect the human faces. Skin color
is different from other colors, and its characteristics
do not change with respect to changes in pose and
occlusion. Because of this, skin color detection is
often the first step of a knowledge based face
detection algorithm.

Comparison between Image based &
Knowledge based methods
Image Based Knowledge Based
Phases of the
systems
The system usually contains two
phases a. Training phase b.
Detection phase
The system usually contains two
phases a. Skin Detection b.Face
detection from the segmented skin
Desigining
Complexity
These systems are more complex as
this involved designing neural
networks to predict the face.
These systems are comparitively
simpler than Image based
techniques
Speed Computationally complex Usually faster than Image based
techniques.
Performance Performance Depends on the
training or learning images
Performance depends on the
criterion used to detect skin, face,
facial features etc

Open Issues
1. There can be insufficient luminance in the image i.e. the image might be too
dark. Skin color cannot be detected in these cases and hence most of the
algorithms fail. And there is not enough difference between the pixels
belonging to the facial region of the image and those belonging to non facial
region of the image.
2. Cases where the face is entirely covered with beard and the eyes are hidden
behind sunglasses have not yet been addressed. In that case neither the skin
detection will work nor will the facial feature detection work.
3. If the background is of skin color and there is a face in front of that where
there is hardly any difference between the skin color of the face and the
background there are no algorithm which can detect it. Though the facial
features might be detected but the exact face will be almost impossible to
detect.
4. Also if the face is rotated to such an extent that the nose cannot be spotted
anymore, then also it would be impossible to identify the facial region with
the existing systems.

Face detection

More Related Content

What's hot

Viewers also liked

Similar to Face detection

Face detection