A CNN-Based Approach for Room Number Detection Using Drone in Indoor Environment

The revolutionary development of drone technology has significantly transformed various sectors such as security, construction, and transportation. By using computer vision and artificial intelligence, drones can now autonomously execute tasks that previously relied on human intervention, resembling intelligent robots. In this evolving landscape, drones have emerged as promising solutions, particularly for indoor applications, due to their safety and efficiency advantages. Despite these advancements, challenges persist, especially in ineffective navigation and airspace management within indoor environments. To address this gap, this current study emphasizes on utilizing a convolutional neural network (CNN) for room number detection within indoor drone systems, which is essential for ensuring precise product delivery. The drone is equipped with a Pixhawk flight controller and a Raspberry Pi for image capture and processing using a CNN model implemented in Keras based on the collected data. Our CNN model has achieved a recognition accuracy of 92.4%, outperforming conventional models in recall and F1 Score, highlighting the effectiveness of CNN-based solutions for advancing indoor drone operations and improving delivery services. The integration of image processing into real-time recognition systems mounted on drones holds immense potential for enhancing delivery efficiency and accuracy. This research demonstrates the transformative capabilities of CNN-based solutions in advancing indoor drone operations, promising more reliable and efficient services in the future.