The study of meteors offers vital insights into the origin, evolution, and dynamics of celestial bodies within our solar system. The collection of meteor trajectories and identification of potential meteorite impact sites provides direct evidence of these astronomical phenomena. With meteors entering Earth’s atmosphere from random directions at unpredictable times, combined with the widespread availability of commercial cameras in surveillance networks, smartphones, and amateur astronomical equipment, the development of systematic data collection and analysis methods has become essential. This paper introduces the HuoLiuXing Project (meaning “Fireball Project” in Chinese), an integrated platform that combines artificial intelligence and digital twin for meteor data collection and management. The project implements multiple methodologies for observational data collection and employs convolutional neural networks for reliable meteor identification. To enhance scientific contributions, we establish digital twins of camera systems, complete with detailed installation and calibration protocols for various equipment types. These digital twins also account for measurement uncertainties and incorporate prior information into analyses. Additionally, we present the digital twin of the entire HuoLiuXing system, which calculates three-dimensional probability distributions of meteor trajectories and potential meteorite impact locations according to observation data from multiple cameras. Experimental validation confirms that our digital twin approach effectively and robustly estimates meteor trajectories and precisely determines possible meteorite landing sites. Since its deployment in 2022, the HuoLiuXing Project has collected over 1400 valid meteor reports from various cameras deployed across China. All scientific data generated through this initiative are made publicly available through the PaperData Repository at https://nadc.china-vo.org/res/r101631/.