SREENIVASA INSTITUTE OF TECHNOLOGY AND MANAGEMENT

STUDIES(AUTONOMUS)
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

TRAFFIC SIGN DETECTION FOR AUTOMATIC SELF DRIVING CAR SYSTEM USING DEEP LEARNING

UNDER THE GUIDANCE OF:

Presented by:
E.GNANA SAI (20751A0548)
E.THARUN(20751A0551)
G.NETHAJI(20751A0560)
G.TEJESHWARAO(20751A0562)
 Abstract:
 The use of deep learning models for traffic sign detection has become increasingly popular in
recent years, particularly in the field of autonomous driving.

 In this approach, a dataset of traffic sign images is collected and preprocessed, and a deep
learning model, such as a convolutional neural network (CNN), is trained on the data.

 The model is then evaluated on a test set and deployed in a real-world scenario. By using deep
learning models, traffic signs can be detected and recognized with high accuracy and speed,
making them an ideal solution for real-time applications.

 Overall, traffic sign detection using deep learning models is a challenging but promising area of
research that has the potential to improve road safety and facilitate the development of
autonomous driving technologies.
Introduction :
 In the past decade the automobile industry has made a shift towards intelligent vehicles equipped
with driving assistance systems. Recently the automobile industry has introduced vision systems in
their high end cars. Examples are the BMW 7er, Mercedes S-Class, Audi A8
 One of the basic use cases of a traffic-sign recognition system is for speed limits and extracted
information is display it in the dashboard of the car to alert the driver about the road sign
 This not only ensures road safety, but also allows the driver to be at little more ease while driving
on tricky or new roads.
 Deep learning models, such as convolutional neural networks (CNNs), are capable of
automatically learning features from raw data, making them well-suited for tasks such as traffic
sign detection.

 By training a CNN on a large dataset of traffic sign images, the model can learn to recognize
traffic signs with high accuracy and speed, even in complex and challenging environments
Existing System
 The existing system of traffic sign detection includes traditional computer vision techniques, such
as color segmentation, edge detection, and template matching.
 By using Logistic regression algorithm, the code made on.
 It is one the most administered and managed machine learning algorithm.
 Based on inputs i.e., images, the algorithm predicts the real valued output.
 Traffic sign prediction is based on these inputs. There are many advantages of traffic management,
the very important being reduction in accident and in time consumption.
 Drawbacks:
 Accuracy is low
 Time consumption is high
 Due to low accuracy, not reliable
Proposed System

• Matrix called "Activation Map" or "Feature Map". The output layer is madeup of several
convolutional layers that extract features from the image . CNN can be optimized with the
help of hyper parameter optimization. It finds hyper parameters of a given deep learning
algorithm that deliver the best performance as measured on a validation set. Hyper
parameters must be set before the learning process can begin. The learning rate and the
number of units in a dense layer are provided by it.

Convolutional Neural Network Architecture :

 This layer is major building block in convolution process. It performs convolution operation to identify various
features from given image. It basically scans entire pixel grid and perform dot product. Filter or kernel is nothing
but a feature from multiple features which we want to identify from input image.

2. Pooling Layer

 This layer is used for down sampling of the features. It reduces dimensionality of large image but still retains
important features. It helps to reduce amount of computation and weights. One can choose Max pooling or
Average pooling depending on requirement. Max pooling takes maximum value from feature map while average
takes average

3.Activation Function

 This layer introduce non linear properties to network. It helps in making decision about which information should
be processed further and which not. Weighted sum of input becomes input signal to activation function to give
one output signal.This step is crucial because without activation function output signal would be simple linear
function which has limited complex learning capabilities.
4.Flattening Layer

 The output of the pooling layer is in the form of a 3D feature map, and we need to transfer data
to the fully connected layer in the form of a 1D feature map. As a result , this layer transforms a 3*3
matrix to a one-dimensional list.

5. Fully connected Layer

 Actual classification happens in this layer. It takes end result of convolution or polling layer by flattened layer and
reaches a classification decision. Here every input is connected to every output by weights .It combines the
features into more attributes that better predicts the classes
System Requirement:
Software Requirements
 IDE : Anaconda , Jupyter
 Programming Language : Python

Hardware Requirements
 PROCESSOR : Intel i5(min)
 RAM : 4 GB DD RAM
 HARD DISK : 250 GB