Vehicle and Building Detection using Convolutional Neural Network for Drone Images

Exploring advanced neural networks, our study concentrates on utilizing deep learning for precise detection of vehicles and buildings. Our aim is to devise an intelligent system capable of accurately discerning objects across diverse environments. Through extensive training on diverse datasets, we ensure the adaptability and reliability of our system across varied scenarios. Leveraging the unique capabilities of neural networks, such as feature extraction and pattern recognition, we optimize our system’s efficiency and accuracy in object detection. Our objective is to streamline usability while minimizing the requirement for sophisticated equipment like drones. By incorporating cutting-edge neural network techniques, our endeavor seeks to redefine the paradigm of object detection, offering practical solutions for urban planning, traffic management, and infrastructure monitoring. We present two models: one employing Convolutional Neural Networks (CNN) with an accuracy of 97.25%, and another based on You Only Look Once (YOLO) V5 architecture achieving 94.26% accuracy. Both models undergo extensive training on diverse datasets to enhance their proficiency in detecting ’bicycle’, ’2-wheeler’, ’3-wheeler’, ’4-wheeler’, and ’buildings’ from drone images.