International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 09 Issue: 05 | May 2022 www.irjet.net p-ISSN: 2395-0072 © 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 478 Lane Detection and Traffic Sign Recognition using OpenCV and Deep Learning for Autonomous Vehicles Birali Prasanthi 1 , Kyathi Rao Kantheti 2 1 Student, Bachelors in CSE, Mahatma Gandhi Institute of Technology, Hyderabad, Telangana, India 2 Senior Assistant Professor, Dept. of Computer Science and Engineering, Mahatma Gandhi Institute of Technology, Hyderabad, Telangana, India ---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract –By the means of automation, number of car crashes on road can be reduced. Testing of autonomous vehicles on public roads can be done on the public roads of the US. Major benefits of automated vehicles include a 90% reduction in traffic deaths, a 60% reduction in harmful emissions, a 40% reduction in travel time, and a 500% increase in lane capacity. Autonomous vehicles are expected to be safer. "Over 90% of accidents today are caused by driver error," said Professor Robert W.Peterson. Autonomous cars are designed for the elimination of traffic created by stop-and-go behavior, according to research done at the University of Illinois. This will be helpful in saving the time of people and as well decreases the time their cars are on the roads, which will reduce the emission of harmful gases from the vehicles. A very close part of driver assistant systems is lane detection. Lane detection refers to the process of tracing white markings on the road, capturing and processing images using a camera mounted in front of the car, and this is done using the OpenCV library. Safety driving also involves recognition of traffic signs as a major part. Promising results have been presented by the CNN (convolutional neural networks). Key Words: CNN, Lane detection, Traffic sign recognition, OpenCV, Autonomous Vehicles. 1. INTRODUCTION The white markings on the road parallel to its direction, usually two in number are spoken as lanes. one of the major elements for a vision-based driver assistance system is lane detection and its tracking. The previous method was accustomed to detecting the obstacle on the road and Therefore the distance of the obstacle with regard to vehicles is still as lanes structured road. In an advanced driver assistance system (ADAS), recognition of traffic signs is incredibly important for safe driving. Promising results have been presented by convolutional neural networks (CNNs) recently. During this work, a strong model supported VGG network by adding batch normalization operation is proposed. To reduce the overfitting of the model, dropout is also used. With the help of the dataset imbalance data augmentation is performed. Then, to boost the images, Contrast limited adaptive histogram equalization (CLAHE) and normalization are performed. The performance of the model is evaluated using various performance metrics such as confusion matrix, precision, recall on German traffic sign recognition benchmark (GTSRB) dataset. According to the Experiments results, the proposed model reaches a state-of- art accuracy of 99.33 % and surpasses the best human performance of 98.84 % 1.1 LITERATURE SURVEY FOR LANE DETECTION During the literature review, it was discovered that the majority of the existing literature has neglected one or more of the following:- 1) According to the survey, the present methods provide good precision for high-quality photographs, although can be a little sloppy at times. Bad outcomes due to poor environmental circumstances such as fog, haze, smog, Noise, dust, and so on 2) The majority of present approaches are better suited to straight lanes. However, they perform poorly on curved roadways. 3) The majority of lane detection systems are based on industry standards. Hough transform can be tweaked to improve the results even more accurate. 1.2 LITERATURE SURVEY FOR TRAFFIC SIGN RECOGNITION In 1987, Akatsuka and Imai conducted the first traffic sign recognition research, attempting to create a very basic traffic sign recognition system. A system capable of recognizing traffic signs on its own and providing aid to drivers by informing them of the presence of a certain restriction or danger, such as speeding or construction work. It may be used to detect and recognize specific traffic signs automatically. A traffic sign recognition system's operation is often separated into two parts:  Detection and  Classification.