Enhanced Yolov5 Deep Learning Technique for Multi-Object Detection in Autonomous Vehicle in Extreme Weather Condition
Year: 2024
Type: article
Abstract: Autonomous vehicle requires more accurate object detection and object classification for the real-world environment. Detection and classification of objects in the stable environment using deep learning framework yields the very efficient result. But in real situation due to rapid changes in the real environment object detection in harsh condition fails due to the various challenges in it. In the proposed work considered two types of dataset image [KITTI] and video [BDD] with nine various environment condition -normal, night, rain, rain with night, low light, high illumination, cluttered environment, FOG, high speed. Applied enhanced robust deep learning algorithm YOLOV5 to all the above conditions and results are compared with the accuracy with the previous work. From the enhanced Yolov5 resulted in 98.7%, 76%, 76%,75%, 71% in normal, rainy, haze, high illumination, night weather condition respective. Proposed model is giving significant improvement in accuracy, precision, recall than the previous work in the extreme weather condition.
Author: Saira Banu Atham., Ranjitha P.,
Institution: Presidency University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7217339304