This project aims to achieve high-fidelity 3D reconstruction of target objects using drones for autonomous image collection and processing through an Instant Neural Radiance Field (Instant-NGP). Motivated by the need for precise, efficient data collection in challenging environments, the project seeks to optimize imaging for 3D reconstruction while minimizing human intervention. The methodology involves training a detection model to autonomously identify and track targets, with the drone capturing images from optimal viewpoints. The collected data is processed using COLMAP for structure-from-motion to estimate camera parameters, which are then utilized to train an optimized Instant-NGP model, ensuring high-quality 3D reconstructions.In the future, incorporating sensor fusion and hybrid methods combining structure-from-motion and sensor-based pose estimation could enable real-time, accurate 3D reconstruction, further enhancing the drone's autonomous capabilities.