Jul 2026· 2026 6th International Conference on Inventive Computation and Information Technologies (ICICIT)· pp. 158-163· 0 citations· 16 references
Abstract
The design of a low-cost driver assistance system (DAS) using monocular camera input and artificial intelligence to enhance road awareness consists of using low-cost sensors instead of costly configurable sensors used in typical systems. The hybrid perception architecture of this system incorporates deep learning (via optimized YOLOv8) and traditional computer vision techniques to achieve high accuracy in detecting vehicles and pedestrians, which is consistent regardless of traffic conditions. Additionally, the hybrid lane detection algorithm combines edge-filtering techniques with geometric models to allow for lane detection in low-light or poorly marked lane conditions. Also, the development of a modular processing pipeline allows for real-time video preprocessing, feature extraction and risk assessment, therefore requiring less computational resources than standard DAS systems. Finally, testing showed that this DAS system consistently performs in real-time and achieves an acceptable degree of accuracy, irrespective of environmental conditions. The DAS system provides a common structure for a variety of vision techniques and can be scaled and constructed for a lower cost than most current DAS solutions, thereby facilitating the development of intelligent transportation systems and increasing access to transportation technology.
Vehicle recognition is a critical component of traffic analysis and the progress of advanced transportation systems, underscoring the importance of automated, real-time methods that reduce the need for manual observation. While the field has seen notable innovations in deep learning-centric detection technologies, many of these approaches require considerable computational strength and are not well-suited for real-time application in resource-constrained environments. In response to this limitation, the present study introduces a streamlined vehicle detection framework that combines background subtraction for motion-oriented foreground extraction with a Haar cascade classifier for object identification in traffic video sequences. The system is evaluated using real-world highway traffic recordings under different illumination conditions, including both day and night scenarios. The experiment's findings show that the system achieves an overall accuracy of 82.08%, with a precision of 85.33%, a recall of 66.67%, and an F1-score of 74.86%. The system also demonstrates consistent performance across different lighting conditions. These findings indicate a trade-off between detection accuracy and computational efficiency, where the proposed approach prioritizes practical deployment feasibility. Overall, the results suggest that classical computer vision techniques remain viable alternatives for real-time traffic monitoring in environments with limited computational resources.
Ni Gusti Ayu Dasriani, Anthony Anggrawan, Khasnur Hidjah et al.· International Journal of Inf...· 0 citations
The study concluded that autonomous car systems have strong potential to revolutionize transportation systems by improving safety and efficiency, although further improvements are required to enhance robustness, real-time adaptability, and user trust in fully autonomous driving environments.
Harold Orji, Ohaeri Ignatius, Udoudom Emem Etim et al.· INTERNATIONAL JOURNAL OF MAT...· 0 citations
Ensuring road safety requires timely and accurate detection of lanes and roadside traffic signs, which are essential for assisting drivers and reducing accident risks. However, robust simultaneous detection remains challenging due to variations in illumination, occlusions, complex backgrounds, and diverse road conditions. This research introduces a novel vision-based system for lane detection and roadside traffic sign recognition using advanced artificial neural network architectures. In the pre-processing stage, Contrast Limited Adaptive Histogram Equalization (CLAHE) is applied to enhance image visibility under varying lighting conditions. For classification, a Self-Artificial Attentive Gated Neuro-Recurrent Unit (SAAG-NRU) is proposed, a novel architecture integrating deep artificial neural network (ANN) layers, Gated Recurrent Units (GRU), and a self-attention mechanism, enabling sequential refinement of features and prioritization of the most safety-relevant cues. In addition, YOLOv11 was utilized to train and detect traffic signs exclusively. For lane detection, a lane mask-based approach is implemented to accurately segment and highlight lane boundaries. The trained model was integrated into a Tkinter-based Graphical User Interface (GUI) to visualize real-time detection outputs effectively. Implemented using Python-based tools, the framework is validated against benchmark lane and traffic signs, demonstrating better performance compared to existing systems in terms of lane detection, achieving 91.11% precision, and traffic sign detection, attaining 96.88% precision. Therefore, the proposed system delivers fast, accurate, and robust simultaneous lane and traffic sign detection, significantly improving real-time road safety and driver assistance.
Viraj Sonawane, B. Agarkar, Sachin Chaudhari· International Journal of Adv...· 0 citations
In an autonomous driving vehicle, the onboard perception system is essential for understanding a dynamic, complex road scene in real time, which is fundamental to autonomous driving. Perception tasks, such as object detection, classification of traffic signs, lane detection, etc., have seen significant progress in autonomous vehicles. One of the major practical challenges is designing an integrated, computationally efficient system capable of supporting multiple perception tasks without compromising detection performance or increasing the hardware cost of embedded automotive platforms. The proposed work introduces an integrated module-based Unified Advanced Driver Assistance System (ADAS) architecture. The first is a Traffic Sign Recognition (TSR) subsystem using the lightweight YOLOv8n backbone for detecting and classifying regulatory, warning, and informational traffic signs under a variety of illumination and environmental conditions. The second module focuses on the Forward Collision Warning (FCW) subsystem, which is trained to detect vehicles, pedestrians, bicyclists, and traffic control signals in dense urban and highway situations. The third module is a novel bimodal distillation fusion module, which bundles the knowledge from both perception streams into a single student model efficient inference pipeline using the principles of knowledge distillation, compressing the knowledge of two specialist detectors into an efficient student model. The architecture has been trained and tested with domain-specific datasets that include realistic noise on the weather, illumination, and complexity. For all the major object and sign categories, significant quantitative evaluation, such as the Precision-Recall curve, the F1-confidence analysis, the confusion matrix, and the Mean Average Precision (mAP) measurement, reveals satisfactory detection performance. Achieved with the TSR module, the mAP@0.5 score is approximately 0.994, and for primary vehicle classes, the mAP@0.5 score is above 0.81 with the FCW module. In addition to the static regulatory information used in fusion, the scenario utilizes dynamic information from the environment, significantly increasing the completeness of the result, which creates a flatter confidence distribution for subsequent ADAS decision logic. The proposed lightweight ADAS architecture supports Sustainable Development Goal (SDG) 9 (Industry, Innovation and Infrastructure) by enabling computationally efficient AI-based intelligent transportation technologies, while also contributing to SDG 11 (Sustainable Cities and Communities) through improved road safety and intelligent mobility.
M. Maddan, N. Sankkarshana, M. Gayathri et al.· Frontiers in Future Transpor...· 0 citations
Accurate detection of lane markings is one of the primary tasks for Advanced Driver Assistance Systems (ADAS) and autonomous vehicles. State-of-the-art lane marking detection algorithms employ object-detection methods based on Convolutional Neural Networks (CNN) and have been widely recognised for yielding accurate detection performance so far. However, pitfalls abound, especially under adverse lighting conditions. Glaring lights or starbursts can create geometric patterns that resemble lane markings, thereby leading to inaccurate detection. To address these challenges, this study presents a novel Vertical Noise Mitigation Module (VNMM) embedded in a hybrid end-to-end CNN architecture that minimises noise interference in the image, thereby increasing the robustness of lane marking detection. VNMM reduces image-related noise and distortion to improve the detection accuracy under adverse lighting conditions. It approaches the problem in a column-oriented classification manner and extracts the features column-wise, since lane markings predominantly appear vertically in an image. The VNMM reduces lateral interference spatially by using channel dimensions to capture noise that is oriented diagonally or laterally. The proposed method is evaluated using a dataset that was specifically collected for this purpose, as well as two widely recognised lane marking datasets: CULane and TuSimple. The method’s effectiveness under adverse lighting conditions is highlighted by the improved accuracy compared with other state-of-the-art techniques in the experimental results. Despite the challenging scenes within the CULane dataset, particularly in crowded scenarios, this approach yielded a lane marking detection accuracy of 81.2%, compared with the previously reported accuracy of 80.6%. Thus, this study contributes significantly to the field of lane marking detection, especially under adverse lighting conditions.
Nima Zarbakht, J. Zou, G. Fang· Multimedia tools and applica...· 0 citations
Current road safety systems face challenges in detecting curved roads and vehicles under varying lighting and weather conditions, leading to lane departure and collision risks. To address this, we propose CurvLaneNet-YOLO, a deep learning framework based on YOLOv8 that simultaneously detects road curvature and vehicles in lanes. The system integrates a parallel polynomial lane detection head into the YOLOv8 architecture, enabling real-time curvature estimation alongside vehicle detection. This research develops an AI-enabled real-time monitoring system to substantially improve road safety. The system makes use of the enhanced capabilities and features of the YOLOv8 model, which involves data preparation, model training, extensive testing, and data augmentation to guarantee model precision. Main goals are estimation of road curvature in real-time, high-performance processing and better assist for the driver. To assess the research, the Cars sample from KITTI dataset on Kaggle, which includes 7,481 images of 640 × 640 resolution, were used. The research achieves an inference latency of 15.4 ± 0.3 ms/image on the test hardware, with mAP@0.5 = 0.9373 and mAP@0.5:0.95 = 0.9217, indicating significant improvements over existing work. This research demonstrates the feasibility of using deep learning techniques for vehicle detection and road curvature estimation in real-time. The results presented in this study show promising potential to integrate into a driver-assistance system, although the data used here is limited to a proof-of-concept validation on the KITTI dataset. The suggested solution is competitive in terms of accuracy and inference time on embedded hardware, providing a potential roadmap for its application in real-world scenarios within the context of intelligent transportation systems.
Amit Pimpalkar, Pranali Dandekar, Harika Vanam et al.· Scientific Reports· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
