This document discusses traffic sign detection and classification. It outlines challenges like variable visual appearances and conditions. The goal is high accuracy recognition in real worlds. It explores feature extraction methods like raw pixels, color histograms, and HOG. Classification algorithms tested include MLP, KNN, SVM, random forest, and CNN using the German Traffic Sign Recognition Benchmark dataset. HOG performed best, and MLP achieved highest accuracy at 98%. The conclusion is HOG is most efficient for extraction and MLP performs best for classification.
2. The problem
Problem statement
Classification of German
Traffic Sign Recognition
Benchmark dataset.
Challenges
Wide variability in visual
appearance.
Illumination , weather
condition, partial
occlusions
Goal
High accuracy in
recognizing signs in real
world.
4. Where can I get
data?
Link
http://benchmark.ini.rub.de/?section=
gtsrb&subsection=news
5. Solution
Proposed Plans:
1. Feature Extraction Methods:
a. Raw pixels
b. Color histograms
c. Histogram of Gradients(HoG)
2. Classification Algorithms:
a. Multi Layer Perceptron (MLP)
b. K Nearest Neighbours (K-NN)
c. Support Vector Machines (SVM)
d. Random Forest
e. Convolutional Neural Network (CNN): LeNet
7. Supervised Classification
Procedures for supervised learning technique, showing the overall training and classification procedures
for (n) classes
Image Credit: http://www.jpathinformatics.org
8. Feature Engineering
Image Preprocessing
1. Input : Image(.ppm)
2. Preprocessing of an image
will enhance the properties
of image like brightness, hue,
contrast and saturation.
Feature Extraction
1. Apply different feature
extraction algorithms like
HOG, Color histograms and
pass features to next stage.
2. Raw image also can be
given to the next stage.
Learning Algorithm
Apply different classifiers
like MLP, KNN, SVM,
Random Forest, CNN.
11. Raw Pixel
In raw pixel extraction method, the
pixel coordinates of image is used
and this information is compared with
other images.
12. Color Histogram
A color histogram is a representation
of the distribution of colors in an
image.
13. Histogram of
oriented gradients
(HOG)
Object detection method
The technique counts occurrences of
gradient orientation in localized
portions of an image.
Image Credit: hog features
15. Multilayer
perceptron (MLP)
A multilayer perceptron(MLP) is a
class of feedforward artificial neural
network. It consists of at least three
layer of nodes. Every node is a neuron
which uses a nonlinear activation
function.
16. K-Nearest
Neighbors(KNN)
In KNN, an object is classified by a
majority vote of its neighbors, with the
object being assigned to the class
most common among its k nearest
neighbors.
17. Support Vector
Machine (SVM)
A SVM constructs a hyperplane or set
of hyperplanes in high dimensional
space that has the largest distance to
the nearest training data point of any
class which leads to good separation.
18. Random Forests
Random Forests are an ensemble
learning method that operate by
constructing a multitude of decision
trees at training time and outputting
the class that is the mode of the
classes of the individual trees.
19. Convolutional
Neural Network
(CNN)
A CNN is a class of deep, feedforward
artificial neural networks in which
hidden layers consist of convolutional
layers, pooling layers, fully connected
and normalization layers.
20. Results
Models HOG Color Histogram Raw Pixel
KNN 96% 97% 92%
MLP 98% 78% 70%
SVM 97% 75% 66%
Random Forest 88% 50% 45%
CNN 48% 54% 91.5%
21. Conclusion
HOG was the most efficient feature extraction method among all the
extraction methods used and the maximum efficiency was obtained
using multi layer perceptron.
22. Reference
1. Man vs. computer: Benchmarking machine learning algorithms for
traffic sign recognition, IJCNN 2013
2. Traffic Sign Recognition with Multi-Scale Convolutional Networks,
IJCNN 2011
3. Traffic Sign Recognition using scale invariant feature transform and
svm