Integrating IoT and honey badger algorithm based ensemble learning for accurate vehicle detection and classification

Abstract

The fusion of Internet of Things (IoT) and deep learning (DL) methods has proven valuable in automating vehicle detection and classification tasks on remote sensing images (RSI). This technology has broad applications, including traffic monitoring, urban planning, and transportation management. Recent advancements have demonstrated the efficacy of DL models like convolutional neural networks (CNN) in RSI classification tasks. In this aspect, this study proposes a novel honey badger optimization algorithm with an ensemble learning-based vehicle detection and classification (HBOAEL-VDC) technique. The purpose of the study is to design ensemble DL models for accurate vehicle identification and classification processes. To accomplish this, the HBOAEL-VDC technique makes use of an improved RetinaNet model for the detection of objects, i.e., vehicles on the RSI. Moreover, the classification of detected vehicles takes place using the ensemble learning process, comprising three DL models, namely gated recurrent unit (GRU), long short-term memory (LSTM), and bidirectional long short-term memory (BiLSTM). Furthermore, the HBOA-based parameter tuning process gets carried out to adjust the hyperparameter values of the DL models and thereby improve the classification results. The simulation outcome of the HBOAEL-VDC approach is tested on benchmark RSI databases. The experimentation outcomes reported the enhanced vehicle classification performance of the HBOAEL-VDC approach over other recent DL models.

Keywords

• Smart environment
• Internet of Things
• Ensemble learning
• Vehicle detection