Issuu

International Research Journal of Engineering and Technology (IRJET)

e-ISSN: 2395-0056

Volume: 11 Issue: 10 | Oct 2024

p-ISSN: 2395-0072

www.irjet.net

IMAGE RECOGNITION AUTONOMOUS VEHICLES USING DEEP LEARNING Radhika D1, Devadharshini S2, Kiruthiga Sri K3, Logeshwari N4, Anisha A5 1

Assistant Professor, Vivekanandha College of Engineering for Women, Tiruchengode, Tamilnadu (India), 2,3,4,5 Student, Vivekanandha, College of Engineering for Women, Tiruchengode, Tamilnadu (India). ---------------------------------------------------------------------------***-------------------------------------------------------------------------Abstract— - The rapid advancement of autonomous driving

these challenges and revolutionizing the field of computer vision. Deep learning algorithms, inspired by the structure and function of the human brain, have demonstrated remarkable capabilities in learning complex patterns and features directly from raw data, without the need for explicit feature engineering. This has led to significant advancements in various domains, including computer vision, natural language processing, and autonomous driving. In the context of autonomous vehicles, deep learning techniques offer the promise of more robust and adaptive object recognition systems that can effectively handle the variability and complexity of real-world driving environments. One of the key challenges in applying deep learning to object recognition in autonomous vehicles lies in the development of models that can operate in real-time and generalize well to diverse object categories and environmental conditions. The You Only Look Once (YOLO) algorithm, particularly its latest version, YOLOv8, has emerged as a state-of-theart solution for real-time object detection tasks. Known for its high accuracy and efficiency, YOLOv8 excels in detecting multiple objects within a scene with minimal computational overhead, making it an ideal choice for autonomous driving applications. In this research project, we aim to leverage the power of deep learning, specifically the YOLOv8 algorithm, for object recognition in a real-world environment for autonomous vehicles. By providing real-world video streams as input and utilizing the comprehensive COCO (Common Objects in Context) dataset, which encompasses a wide range of object categories, we seek to develop a robust and efficient system for accurately detecting and classifying objects encountered by autonomous vehicles on the road.

technology has necessitated the development of more sophisticated methods to ensure the safety and effectiveness of these systems. This project report focuses on the implementation and evaluation of an autonomous vehicle using the deep learningbased image recognition system powered by YOLOv8 (You Only Look Once version 8) algorithm. As the latest iteration in the YOLO series, YOLOv8 is acclaimed for its real-time object detection capabilities. It is meticulously designed for high efficiency, achieving an optimal balance between detection of speed and accuracy. YOLOv8 has been trained on an extensive dataset that covers a wide array of driving scenarios, significantly enhancing its ability to recognize various objects on the road, such as vehicles, pedestrians, and cyclists. The report provides an in-depth analysis of the data preparation procedures, the architectural design of the model, and comprehensive training methodology. This process includes model optimization, validation strategies to ensure the model performance in real-world autonomous driving scenarios. The project emphasizes the scalability of YOLOv8, demonstrating its compatibility with a wide spectrum of computational platforms and its adaptability to various vehicle models, ranging from luxury to economy classes, ensuring versatile deployment in the automotive industry. Keywords— YOLOv8, driving scenarios, methodology, pedestrians, and cyclists, pedestrians, and cyclists.

INDRODUCTION

The advent of autonomous vehicles heralds a transformative shift in transportation, promising safer and more efficient journeys on our roads. Object recognition in a real-world environment represents a critical component of autonomous vehicle technology, enabling vehicles to identify and classify various objects encountered on the road, such as vehicles, pedestrians, cyclists, traffic signs, and obstacles. Traditional approaches to object recognition in autonomous vehicles have relied on a combination of sensor data, such as LiDAR, radar, and cameras, along with handcrafted features and rule- based algorithms. While these methods have made significant strides in enhancing vehicle perception, they are often limited by their inability to generalize to diverse and complex real- world scenarios. The emergence of deep learning methodologies has opened up new avenues for addressing

Impact Factor value: 8.315

II. LITERATURE REVIEW 1. 1) You Only Look Once: Unified, Real-Time Object Detection • Publication Year 2015 • Authors: Joseph Redmon, Santosh Divvala, Ross Girshick, Ali Farhadi • Summary: YOLO introduced a groundbreaking algorithm for real- time object detection by dividing images into a grid, predicting bounding boxes, and class probabilities directly. This approach ensured high accuracy and speed, making it ideal for applications like autonomous vehicles, where timely and precise object detection is crucial for safe navigation and decisionmaking.

ISO 9001:2008 Certified Journal

Page 394