Multi-Pedestrian Detection using Hybrid ML Algorithms for Autonomous Vehicles DOI
Satyaki Mukherjee, Tanya Sharma,

Anshika Singh

и другие.

Опубликована: Ноя. 23, 2023

Forecasting vulnerable road user behavior is a prerequisite for the real-world implementation of Autonomous Driving Systems (ADS). The purpose pedestrian crossing should be detected instantaneously, particularly while driving in towns. This paper aims to detect multiple pedestrians and other automobiles specifically on Indian Roads real-time. Recent research suggests that vision-based models utilizing deep neural networks are useful this purpose. For we aim develop an end-to-end intention detection architecture works well both during day at night. main approach project based bounding boxes object identification. using various learning techniques like YOLOv3, Darknet-53 YOLOv7.

Язык: Английский

Fast and accurate object detector for autonomous driving based on improved YOLOv5 DOI Creative Commons
Xiang Jia, Ying Tong,

Hongming Qiao

и другие.

Scientific Reports, Год журнала: 2023, Номер 13(1)

Опубликована: Июнь 15, 2023

Abstract Autonomous driving is an important branch of artificial intelligence, and real-time accurate object detection key to ensuring the safe stable operation autonomous vehicles. To this end, paper proposes a fast detector for based on improved YOLOv5. First, YOLOv5 algorithm by using structural re-parameterization (Rep), enhancing accuracy speed model through training-inference decoupling. Additionally, neural architecture search method introduced cut redundant branches in multi-branch module during training phase, which ameliorates efficiency accuracy. Finally, small layer added network coordinate attention mechanism all layers improve recognition rate vehicles pedestrians. The experimental results show that proposed KITTI dataset reaches 96.1%, FPS 202, superior many current mainstream algorithms effectively improves performance unmanned detection.

Язык: Английский

Процитировано

56
Low-cost autonomous car level 2: Design and implementation for conventional vehicles DOI Creative Commons

Mohammad S. Mohammed,

Ali M. Abduljabar,

Mustafa M. Faisal

и другие.

Results in Engineering, Год журнала: 2023, Номер 17, С. 100969 - 100969

Опубликована: Фев. 28, 2023

Modern cars are equipped with autonomous systems to assist the driver and improve driving experience. Driving system (DAS) is one of most significant components a self-driving vehicle (SDV), used overcome non-autonomous challenges. However, conventional not DAS, high-cost required equip these vehicles DAS. Moreover, design DAS very complex outside industry while it requires going through Electronic Control Unit (ECU), which has high level security. Therefore, basic needs be installed in makes more efficient terms assistance. In this paper, an intelligent presented for real-time prediction steering angle using deep learning (DL) raw dataset collected from real environment. Furthermore, object detection model deployed warn various types objects along corresponding distance measurement based on DL. Outputs DL models fed into control system, Power Steering (EPS). The measured time sensor posted back make automated adjustments accordingly. Real-time tests conducted 2009 Toyota Corolla digital camera capture live video stream, Controller Area Network (CAN-BUS) messages, sensor. performance evaluation proposed indicates assistance when evaluated

Язык: Английский

Процитировано

31
Advances in Facial Expression Recognition: A Survey of Methods, Benchmarks, Models, and Datasets DOI Creative Commons
Thomas Kopalidis, V. Solachidis, Nicholas Vretos

и другие.

Information, Год журнала: 2024, Номер 15(3), С. 135 - 135

Опубликована: Фев. 28, 2024

Recent technological developments have enabled computers to identify and categorize facial expressions determine a person’s emotional state in an image or video. This process, called “Facial Expression Recognition (FER)”, has become one of the most popular research areas computer vision. In recent times, deep FER systems primarily concentrated on addressing two significant challenges: problem overfitting due limited training data availability, presence expression-unrelated variations, including illumination, head pose, resolution, identity bias. this paper, comprehensive survey is provided FER, encompassing algorithms datasets that offer insights into these intrinsic problems. Initially, paper presents detailed timeline showcasing evolution methods expression recognition (FER). illustrates progression development techniques resources used FER. Then, review introduced, basic principles (components such as preprocessing, feature extraction classification, methods, etc.) from pro-deep learning era (traditional using handcrafted features, i.e., SVM HOG, era. Moreover, brief introduction related benchmark (there are categories: controlled environments (lab) uncontrolled (in wild)) evaluate different comparison models. Existing neural networks strategies designed for based static images dynamic sequences, discussed. The remaining challenges corresponding opportunities future directions designing robust also pinpointed.

Язык: Английский

Процитировано

17
An in-depth evaluation of deep learning-enabled adaptive approaches for detecting obstacles using sensor-fused data in autonomous vehicles DOI
Abhishek Thakur, Sudhanshu Mishra

Engineering Applications of Artificial Intelligence, Год журнала: 2024, Номер 133, С. 108550 - 108550

Опубликована: Май 9, 2024

Язык: Английский

Процитировано

16
A Framework for Communicating and Building a Digital Twin Model of the Electric Car DOI Creative Commons
Tomasz Bednarz, A. Baier, Iwona Paprocka

и другие.

Applied Sciences, Год журнала: 2024, Номер 14(5), С. 1776 - 1776

Опубликована: Фев. 22, 2024

The Fourth Industrial Revolution has had a huge impact on manufacturing processes and products. With rapidly growing technology, new solutions are being implemented in the field of digital representations physical product. This approach can provide benefits terms cost testing time savings. In order to test reflect operation an electric car, twin model was designed. paper collects all information standards necessary transform idea into real virtual car. significance study improvement project described. research stand, correlations components (DC AC motors, shaft, wheel car), development prospects presented paper. communication method with stand is also presented. should communicate time, which means obtaining correct output when input changes; motor current, rotational speed DC motor. relation between inputs outputs tested. kinematics car modelled LabVIEW. results obtained compared historic racing data. track modeled based satellite data, taking account changes terrain height, using SG Telemetry Viewer application. parameters engine tuned actual data car’s current achieved then discussed.

Язык: Английский

Процитировано

6
Involvement of Deep Learning for Vision Sensor-Based Autonomous Driving Control: A Review DOI

Abida Khanum,

Chao-Yang Lee, Chu‐Sing Yang

и другие.

IEEE Sensors Journal, Год журнала: 2023, Номер 23(14), С. 15321 - 15341

Опубликована: Июнь 5, 2023

Currently, autonomous vehicles (AVs) have gained considerable research interest in motion planning (MP) to control driving. Deep learning (DL) is a subset of machine motivated through neural networks. This article provides the latest survey on theories and applications DL, reinforcement (RL), deep RL, it summarizes different DL methods. In addition, we present main issues driving (AD) analyze DL-based architectures for decision-making frameworks MP tasks, such as lane assist, following, overtaking, collision avoidance, emergency braking, MP. Furthermore, introduce well-known publicly available datasets collected public roads simulators suitable AD purposes discuss simulator environments, activation functions, libraries output AVs. Moreover, challenges terms hardware software, safety, computational time cost, balanced data, multitask learning, technology issues. Finally, future directions

Язык: Английский

Процитировано

12
Robust autonomous driving control using deep hybrid-learning network under rainy/snown conditions DOI
Chao-Yang Lee,

Abida Khanum,

Tien‐Wen Sung

и другие.

Multimedia Tools and Applications, Год журнала: 2024, Номер unknown

Опубликована: Июнь 13, 2024

Язык: Английский

Процитировано

3
Cutting‐Edge Deep Learning Methods for Image‐Based Object Detection in Autonomous Driving: In‐Depth Survey DOI Creative Commons

Narges Saeedizadeh,

Seyed Mohammad Jafar Jalali, Burhan Khan

и другие.

Expert Systems, Год журнала: 2025, Номер 42(4)

Опубликована: Фев. 27, 2025

ABSTRACT Object detection is a critical aspect of computer vision (CV) applications, especially within autonomous driving systems (AVs), where it fundamental to ensuring safety and reducing traffic accidents. Recent advancements in computational resources have enabled the widespread adoption Deep Learning (DL) techniques, significantly enhancing efficiency accuracy object tasks. However, technology for has yet reach level maturity that guarantees consistent performance, reliability, safety, with several challenges remaining unresolved. This study specifically focuses on 2D image‐based methods, which offer advantages over other modalities, such as cost‐effectiveness ability capture visual features like colour texture are not detectable by LiDAR. We provide comprehensive survey DL‐based strategies detecting vehicles pedestrians using images, analysing both one‐stage two‐stage frameworks. Additionally, we review most commonly used publicly available datasets research highlight their relevance The paper concludes discussing current this domain proposing potential future directions, aiming bridge gap between capabilities models requirements real‐world applications. Comparative tables included facilitate clear understanding different approaches datasets.

Язык: Английский

Процитировано

0
Real-Time Deep Learning Based Safe Autonomous Navigation DOI

Tuhin Dutta,

D Santhosh Reddy,

P. Rajalakshmi

и другие.

Опубликована: Янв. 12, 2024

Язык: Английский

Процитировано

2
Development of an Autonomous Driving Vehicle for Garbage Collection in Residential Areas DOI Creative Commons

Jeong-Won Pyo,

Sang-Hyeon Bae,

Sung-Hyeon Joo

и другие.

Sensors, Год журнала: 2022, Номер 22(23), С. 9094 - 9094

Опубликована: Ноя. 23, 2022

Autonomous driving and its real-world implementation have been among the most actively studied topics in past few years. In recent years, this growth has accelerated by development of advanced deep learning-based data processing technologies. Moreover, large automakers manufacture vehicles that can achieve partially or fully autonomous for on real roads. However, self-driving cars are limited to some areas with multi-lane roads, such as highways, drive urban residential complexes still stage. Among various purposes, paper focused garbage collection areas. Since we set target environment vehicle a complex, there is difference from general vehicle. Therefore, paper, defined ODD, including length, speed, conditions area. addition, recognize vehicle's surroundings respond situations, it equipped sensors additional devices notify outside state operate an emergency. system capable object recognition, lane route planning, manipulation, abnormal situation detection was configured suit hardware way. Finally, performing actual experimental section developed vehicle, confirmed function area works appropriately. would support through experiment work efficiency.

Язык: Английский

Процитировано

7