Color-Based Auto-Rickshaw Detection Using Convolutional Neural Networks for Intelligent Transportation Systems
Authors
Md. Wadud Jahan
(Computer Science and Engineering)
Abstract
Gait recognition is a crucial biometric tool for surveillance and security, but its effectiveness is hindered by factors like clothing, carried obj ects, and environmental conditions. This study introduces an adaptive framework that assigns weights to gait components to counter these issues. Using EfficientNet_B7, a deep learning technique, the framework reduces intra-class variations and improves recognition accuracy. The experimental results show a 98.7% accuracy rate, proving the method's robustness in various conditions. This advancement significantly enhances the reliability of gait recognition systems in real-world applications.
