The intelligent vehicle, as a crucial technology in the intelligent transportation system, is the bearer of a complete integration of several technologies. Although vision-based autonomous driving has showed great promise, there is still the issue of analysing the complex traffic scenario using the data acquired. Recently, self driving has been broken down into many tasks utilising various models, such as object detection and intention identification. In this work, a vision-based system was built to recognise and identify numerous items in the traffic scene, as well as forecast pedestrian intentions. The key contributions of this study are: (1) A fine-tuned Part-Affinity-Field's approach to know pedestrian pose was proposed; (3) Explainable-AI (XAI) technology was used to explain and assist the estimation results in the risk assessment phase; and (4) a large autonomous driving dataset with several subsets for each task was brought in; and (5) an endtoend system having multiple models with high accuracy was developed. The overall parameters of the modified Faster RCNN were decreased by 74%, demonstrating that it meets the real-time capabilities. Furthermore, when compared to the state-of-the-art, the detection precision of the enhanced Faster RCNN improved by 2.6 percent.
Article Details
Unique Paper ID: 155467
Publication Volume & Issue: Volume 9, Issue 1
Page(s): 975 - 978
Article Preview & Download
Share This Article
Join our RMS
Conference Alert
NCSEM 2024
National Conference on Sustainable Engineering and Management - 2024