Real-Time Vehicle Classification Using LSTM Optimized by Oppositional-Based Wild Horse Optimization
Year: 2024
Type: article
Abstract: Classifying vehicles in real time was necessary to manage and plan road traffic and avoid frequent traffic jams, traffic violations, and fatal traffic accidents.However, detecting vehicles at night presents a significant challenge, requiring the classification algorithm to be tested under diverse conditions, such as rainy weather, cloudy weather, low illumination, and others, which makes identifying vehicles a complicated task.This paper detected and classifiess vehicle through YOLO-v2, ResNet50, and an optimally configured Long Short-Term Memory (LSTM).But figuring out the best hyperparameters by trial and error took longer and was more complicated.The research resolved the computational time and complexity by involving Oppositional-based Wild Horse Optimization (OWHO) techniques to identify the optimal hyperparameters for LSTM.The result showed that the proposed technique was better, with an average accuracy of 97.38% in classifying vehicles, which was better than other techniques.
DOI: 10.18280/ria.380410
Authors: Kendagannaswamy Tejaswi, Ramaiah Krishna Bharathi
Institution: JSS Science and Technology University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4401828113