| Authors | DR. JYOTI. Y. KULKARNI |
|---|---|
| Article Type | Research Article |
| Language | English |
| Journal | North Asian International Research Journal of Multidisciplinary |
| ISSN | 2454-2326 |
| Volume | 11 |
| Issue | 4 |
| Pages | 30-41 |
| Publication Year | 2025 |
| Publication Date | April 01, 2025 |
| DOI URL | https://doiglobal.org/10.2025/NAIRJCMD.019 |
Automatic traffic-sign perception is a fundamental component of advanced driver-assistance systems, autonomous vehicles and intelligent transportation systems. Indian road environments make this task particularly difficult because traffic signs may be small, partially occluded, faded, damaged, visually cluttered or observed under strong illumination and weather variations. This paper presents a deep-learning framework for real-time detection and classification of Indian traffic signs from road images and video streams. The proposed system uses a one-stage YOLO-family detector as the principal real-time architecture and incorporates image normalization, contrast enhancement, data augmentation, multiscale learning, transfer learning and confidence-based non-maximum suppression. The research design emphasizes Indian datasets and real-world conditions rather than relying only on international benchmark signs. Published evidence is used as comparative context: an Indian Refined Mask R-CNN study used 6,480 images containing 7,056 sign instances across 87 categories and reported 97.08% precision; a 2026 study on IRTSD-Datasetv1 describes 5,141 images spanning 37 traffic-sign classes from more than 90 Indian cities and reports an increase in baseline CNN accuracy from 89.3% to 96.7% after preprocessing; and an improved YOLOv4 study introduced an Indian dataset containing more than 3,000 images for unconstrained traffic-sign detection. These published figures are not represented as results of the proposed experiment. The paper defines a reproducible evaluation protocol using precision, recall, F1-score, mean Average Precision, latency and frames per second. It concludes that a lightweight YOLO-based system, trained on diverse Indian road imagery and tested under unseen environmental conditions, offers a practical route toward real-time traffic-sign perception, but safety-critical deployment requires rigorous external validation.
Total Views
Total Downloads
Publication Year
DOI Status
This scholarly article has been successfully registered and assigned a persistent DOI through DOIGLOBAL.
DOI: 10.2025/NAIRJCMD.019
Journal:
North Asian International Research Journal of Multidisciplinary
DOI Provider:
DOIGLOBAL
This scholarly record is permanently registered through DOIGLOBAL DOI Infrastructure and remains accessible using its persistent DOI identifier.