• In this paper the algorithm based on deep convolutional neural network using region
proposal network in Faster R-CNN. The method can detect all seven main categories
Chinese traffic sign.
• They have trained and compared three models VGG16, VGG_CNN_M_1024 and ZF.
• It gives the traffic sign detection rate of their algorithm is above 99%.
• They don’t have accuracy in obtaining ground truth and the class of the sign detected
• Contours Information algorithm is used which is used in Canny edge.
• Hough Transform method is used to find curve in the image.
• For classification neural networks is used.
• Sign can be detected in any light conditions.
• Image compression is not done which increases the processing time.Advantages:
• Averaging Filter, Wiener Filter, Un-sharp Mask Filtering methods are used to pre-process
• When there is a potential danger of collision Automatic Braking System (ABS) and an
audio-visual warning signal is given.
• They have only done until pre-processing, but sign detection, recognition and
classification is not done.Advantage:
• In this paper they used Optical Character Recognition(OCR) method, which is used to
recognize character in the street board.
• GLCD display is interfaced to Raspberry Pi for displaying text.
• Region of Interest (ROI) method is used to detect the specific character in the street
• Only used for detecting characters in the board but not for traffic sign.Advantages:
• In this paper they used Bilateral Chinese Transform(BCT) method, which is based on the
gradient orientation and magnitude of the edge points. It can detect the complete or
broken circle pattern in the image.
• Vertex and Bisector Transform(VBT) method is used, which is like BCT but also used to
calculate the accumulated contribution of all pairs of edge points.
• Color information is used for selecting colors for detecting traffic sign board.
• Neural Network is used for Recognizing and classification.
• Only 84% of accuracy in detecting the sign board.This project aims detection and recognition by recorded video sequence of traffic
signs of car camera. Traffic signs recognition(TSR) is used to specify the traffic signs
by warning the driver and command or prohibited some actions. TSR support driver
to identifying traffic signs by fast computational using the power full algorithms to
improve driving safety and make comfort to driver. Automatic identification of traffic
signs for automatic intelligent drive vehicle is also important. In this paper they uses
OpenCV to recognition of traffic signs modes of technology. Image was extract, detects and
identification of traffic signs by pretreatment techniques like Canny Edge
Detection, Gaussian filter and thresholding.
In this article they have been achieved openCV method to represent the outlines
in the traffic signs.
They use Artificial Neural Networks in recognition of images.
HSV algorithm I is used to recognition of traffic signs in cylindrical shape.
Computational time increases because of using the large image dataset. Compression
techniques is not used to make computationally fast. Erosion and Dilation are not usedABSTRACT:
In this paper presented a positive visual system of real-time traffic signs recognition
of system is by using two cameras, one is equipped with a wide angle lens other with
telephoto lens and, PC on an image processing board of system is first detect traffic
signs using wide angle camera based on color,content and shape of information. Recognizing
traffic signs, it is often necessary knowledge to known about traffic signs
and symbols. Here in this paper they made system to detect only circular shape signs
and they used image processing board to do pattern matching between capture image
with stored images.
They used normalized coo-relation to match pattern of between captured image
and stored image.
For calculating normalized correlation between capture image and image
database they use image processing board has built in function.
Only circular board is detected.
Normalized correlation method may weak for changing in lighting condition BY: HARINI
Road sign is important to ensure to prevents from accidents. Road symbols is shown
statements have different necessary information need to understand by driver. In
this paper they presented an overview of traffic signs board detection and
identification and achieve a program to extract the signs from a original images.It
gives alert message to driver by text messages only. Automatic driver assistance helps
driver to prevent from accidents by sending an alert message, it may helps driver to
reach their destination and also helps to save fuel and money to driver.
The red color traffic signs are recognized and classified using Hough Transform
and correlation technique. Noise Removing, Thinning, Contrasting operations are
implemented. Dilation and erosion are used to improve image quality.
Computation time is more.
They assured about only 72% of accuracy in detecting and recognizing of traffic