International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
Lane-lines identification system using CNN Pratiksha Chame1, Abhishek Jadhav2, Pranjal Deshmukh3, Prof. Supriya Lole4. 1Student at Dept. of MCA, GHRREM, Pune 2Student at Dept. of MCA, GHRREM, Pune 3Student at Dept. of MCA, GHRREM, Pune
4Professor, Dept. of MCA, GHRCEM, Pune, Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In recent times, significant advancements
seamless, uninterrupted assumptions of lane lines in traffic situations, surpassing the capabilities of previously developed networks. Despite advancements, vehicle crashes remain a significant concern in Malaysia and other Asian countries with tens of thousands of fatalities and millions of injuries annually mostly on highways. The United Nations has ranked Malaysia among the nations facing considerable challenges in road safety.
have been made in the realm of road safety due to the alarming increase in accidents with one of the leading causes being driver in- attention. To mitigate the occurrence of accidents and ensure safety it is imperative to explore technological breakthroughs. One such approach involves the utilization of Lane Detection Systems, which operate by identifying lane boundaries on the road and notifying the driver if they deviate from the correct lane marking. Lane detection systems are pivotal components in various advanced transportation systems. However, achieving this objective proves to be challenging due to the diverse road conditions encountered, especially during nighttime or daytime driving. By placing a camera at the front of the vehicle, it captures the road view and detects lane boundaries. Various techniques have been presented for detecting lane markings on the road. In this study, the approach utilized involves partitioning the video image into smaller segments and extracting image characteristics for each segment. These characteristics are subsequently used to detect the lanes on the road. Keywords- lane Detection System, self-driving car assistant, Open CV, Road Accidents, Machine learning.
1.1 Problem statement According to the Topic overview, the Lane Detection System is extremely important and essential to gain control over the rising number of accidents and save people's lives. It is widely known that the lack of advanced features in vehicles and the increasing driver drowsiness caused by heightened stress are endangering many lives. Taking this into account, a solution needs to be designed. The existing lane detection systems are not accurate and fail to notify the user in any way. Additionally, the new system should be less timeconsuming and more effective. Therefore, this proposed system must be well-developed and responsive. 1.2 Existing System
Key Words: Detection System, self-driving car assistant, Open CV, Road Accidents, Machine learning.
The existing lane line identification system relies on Convolutional Neural Networks (CNNs) and preprocessing of raw images to extract lane markings. However, its reliance on annotated datasets may limit its performance in diverse driving conditions, where training examples are lacking. Additionally, CNNs may struggle to generalize to new environments, leading to reduced adaptability in varying lighting, road markings, and weather conditions. Despite multiple stages including data collection, preprocessing, training, and evaluation, the system may face challenges in accurately identifying lane lines across real-world driving scenarios [2].
1. Introduction A Lane Detection System is a critical component of artificial intelligence (AI), machine learning (ML), and computer vision, particularly in intelligent vehicle systems. Its primary objective is to accurately detect lane lines and issue alerts or messages when a vehicle is about to deviate from its lane. Additionally, it incorporates object detection to prevent accidents. Advanced Driving Assistance Systems (ADAS) rely on lane detection to model road lanes and determine the vehicle's position accurately. The development of intelligent vehicles aims to automate driving tasks, enhancing safety conditions. Road detection is important in driving assistance systems as it provides information such as lane structure and vehicle position which helps for drivers to drive safely. Performing lane detection in complex traffic driving scenes presents numerous challenges [1]. A suggestion has been put forward that spatial CNNs can effectively acquire knowledge about the spatial connections between feature maps and the
© 2024, IRJET
|
Impact Factor value: 8.226
2. Literature Survey We have done research from different research papers and we have found what methods are used and drawbacks in the existing system.
|
ISO 9001:2008 Certified Journal
|
Page 1740