TRAFFIC SIGN RECOGNITION USING CONVOLUTIONAL NEURAL NETWORK

OUR PROJECT USES THE CONVOLUTIONAL NEURAL NETWORK TO RECOGNISE TRAFFIC SIGNS IN REAL TIME WITH THE HELP OF TensorFlow AND KERAS . OUR REAL-TIME TRAFFICS SIGN RECOGNITION SYSTEM IS LATER IMPLEMENTED IN THE ADVANCED DRIVER ASSISTANCE SYSTEM AND SELF DRIVING CARS.

• USB Webcam
• PC or Raspberry Pi

Driving a vehicle in the fast-moving world is an hourly urgent action in human life. inattentiveness in obeying the traffic signs and carelessness in seeing the signboard has significantly increased the accident rate on daily basis. It also causes injuries to pedestrians and animals which is not the only use case. The future is going to be an era of autonomous vehicles, without traffic sign identification autonomous vehicles and driver assistance systems cannot work.

The identified solution for avoiding accidents and identifying the traffic sign in autonomous vehicles the state of the art deep learning technology used as a solution to the described problem. In particular, a convolution deep neural network is used as an effective algorithm to recognize the different traffic signs. The algorithm is used to execute the feature extraction and classification. Thus, traffic sign recognition will be an important feature for the futuristic autonomous vehicle and driver assistance system and after successful recognition, the recognized sign will be voice intimated (voice assistance)

BLOCK DIAGRAM:

Convolutional Neural Networks (CNNs) learns multi-level features and classifier in a joint fashion and performs much better than traditional approaches for various image classification and segmentation problems.

1. Convolution Layer :

The primary purpose of Convolution in case of a CNN is to extract features from the input image

• Pooling layer :

Reduces the dimensionality of each feature map but retains the most important information. Pooling can be of different types: Max, Average, Sum etc

Story So Far :

•Together these layers extract the useful features from the images.

•The output from the convolutional and pooling layers represent high-level features of the input image.

WE TRY TO IMPLEMENT IT FOR REAL-TIME VIDEO CAPTURING

AND WE PROVIDE A VOICE ASSISTANCE FEATURE WHICH IS INNOVATIVE SOLUTION TO AVOID ACCIDENTS

• IT IS PATR PROGRAM OF AUTONOMOUS DRIVING VEHICLE
• IT IS A PART OF THE ALGORITHM FOR THE ADVANCED DRIVER ASSISTANCE SYSTEM
• IMPLEMENTED IN AUTOMATED DELIVERY BOTS
• PART OF ALGORITHM IN AUTOMATIC GARBAGE COLLECTION VEHICLE SINCE IT ALSO TRAVELS ON ROAD