Fast Human Detection Using a Novel Boosted Cascading Structure With Meta Stages

1
Fast Human Detection Using a Novel Boosted
Cascading Structure With Meta Stages
Yu-Ting Chen and Chu-Song Chen, Member, IEEE
2
INTRODUCTION

• TECHNIQUES for detecting humans in images have a
  wide variety of applications, such as
• video surveillance,
• smart rooms,
• content-based image/video retrieval,
• and intelligent transportation systems (ITS)?

3
ABSTRACT

• We propose a method that can detect humans in a
  single image based on a novel cascaded structure
• .
• USE intensity-based rectangle features ,
• gradient-based 1-D features,
• real AdaBoost algorithm,
• a novel cascaded structure
• (standard boosted cascade),
• meta-stages?

4
REAL ADABOOST AND FEATURE POOL

• Intensity-Based Features

denotes the rth type rectangle feature
(r110)
and are the Illumination
summations in the white and black regions,
is the feature value in block
The integral-image method is used for fast
evaluation of these features,but the results for
human detection are not satisfactory.so add
discrimination
In each block,a feature value can be calculated
5
REAL ADABOOST AND FEATURE POOL

• Gradient-Based Features
• HOG
• First, the representation is too complex to
  evaluate, resulting in a slow detection speed.
• Second,all the dimensions of a HOG feature
  vector are employed simultaneously, so it is not
  possible to just use some of them to achieve
  efficient detection.
• Third, its computation cost is high since
  it uses a Gaussian-kernel SVM instead of linear
  SVM.
• EOH
• A EOH feature can only characterize one
  orientation at a time, and it is represented by a
  real value.
• Many EOH features (with respect to different
  orientations)can be extracted from an image
  region, but each feature is only 1-D.

6
REAL ADABOOST AND FEATURE POOL

• EOF Features

First The gradient image is calculated from the
original image
by convolving the edge
operator.
7
REAL ADABOOST AND FEATURE POOL
Second To compute the EOH features, the pixel
gradient magnitude m and gradient orientation ?
of each pixel p at location (x,y) in block Bi.
Gx gradients in the horizontal directions Gy
gradients in the vertical directions
The gradient orientation is evenly divided into K
bins over 0 to 180 . The sign of the orientation
is ignored thus,the orientations between 180 to
360 are deemed the same as those between 0 and
180 .
8
REAL ADABOOST AND FEATURE POOL
Third The gradient orientation histograms Ei,k
in each orientation bin K of block Bi are
obtained by summing all the gradient magnitudes
whose orientations belong to bin K in Bi.
Fourth is the feature value of the Kth
( K 1 K) EOH feature in block Bi. e is a small
positive value that avoids the denominator being
zero.
9
REAL ADABOOST AND FEATURE POOL
Gradient-Based ED (edge-density) feature For a
block , an ED (edge-density) feature is defined
as the average gradient magnitude
is the ED feature value in Bi and ai is
the area of Bi
Similar to the rectangle features, the
integral-image method can be employed for fast
evaluation of the ED features.
Combined Feature Pool
r 110, k 1K
10
REAL ADABOOST AND FEATURE POOL

• real AdaBoost algorithm

Given input data z and its feature Value f(z),
the weak learner output h(z)
After selecting T weak classifiers,the strong
classifier of Real AdaBoost can be expressed as
a is a threshold
A high confidence value implies that the input
data is likely to be a positive sample.
11
CASCADING FEED-FORWARD CLASSIFIERS
A.Contains S stages and Ai is referred to as an
AdaBoost classifier in the ith stage. B.In this
cascaded structure, detection windows that do not
contain humans
C.To find an object of unknown position and size
in an image usually involves a brute-force search
of all possible sites and scales in the image.
Since there are usually far more negative windows
than positive windows to detect in an image,
saving on the detection time of the negative
windows increases the overall efficiency of the
object detector.
DSince more difficult negative examples are used
for training in later stages.
In the current stage will not be used in later
stages.
12
CASCADING FEED-FORWARD CLASSIFIERS

To train a cascaded structure, the goals of the
minimum detection rate of positive examples, di,
and the maximum false-acceptance rate of negative
examples,fi , are set for each stage Ai.
13
CASCADING FEED-FORWARD CLASSIFIERS

• Adding Meta-Stages

A and M denote the AdaBoost stages and
meta-stages Meta-stages 2-D space Mi 2-D vector
14
CASCADING FEED-FORWARD CLASSIFIERS

• Meta-Stage Classifier
• we choose the linearSVM(LSVM) as the meta-stage
  classifier because of its high generalization
  ability and efficiency in evaluation.
• ? is the 2-D normal vector of the plane and
• ß is the offset from the origin.
• The confidence value of the meta-stage for data
  is defined as

15
RESULTS

• rectangle features Rec-Cascade (625)
• EOH features EOH-Cascade (584)
• ED features ED-Cascade (2492)
• a combination of rectangle and EOH features
  RecEOH-Cascade(325)
• a combination of them as feature
  RecEOHED-Cascade (310)

16
RESULTS
In our experiments, the maximum FPPW values are
about 10(-3) for most of the cascaded approaches
compared. Since there are far more negative
windows than the positive windows in an image, a
detector shall have a very low false positive
rate (e.g., under 10(-3)), or it might not be
practically useful.
HOG add META than RecEOHED-Cascade MetaCascade-2D
greatest
miss rate versus false positives per window
(FPPW)
17
RESULTS
In our experiments, we consider thefollowing
three forms , 2-D meta classifiers ( n1 2, ni
1 ) 3-D meta classifiers ( n1 3, ni 2 ) 4-D
meta classifiers ( n1 4, ni 3 ) for a 320X240
testing image, the average processing speeds
MetaCascade-2D 8.61 fps (243) MetaCascade-3D
8.52 fps (230) MetaCascade-4D 8.44 fps
(201) HOG-MetaCascade-2D 6.12 fps (516)
18
RESULTS
MetaCascade-2D 8.61 fps MetaCascade-3D 8.52
fps MetaCascade-4D 8.44 fps
HOG-MetaCascade-2D 6.12 fps
HOG-LSVM (0.91 fps)
19
RESULTS
20

• Thank you for your
• listening !
• 2008.10.21