With the continuous development of urban rail transit, the demand and quantity of shield tunnel construction are constantly increasing. Backfill grouting behind the shield tunnel lining is a crucial technology for controlling engineering issues such as ground disturbance and segment misalignment, which is vital for ensuring the safety of shield construction and surface buildings. In order to achieve automation and intelligence in backfill grouting detection, and to address the low efficiency of traditional Ground Penetrating Radar (GPR) manual detection, a Loaded-to-Frame (LTF) equipment capable of rapid automated GPR detection of grouting quality is proposed. The hardware composition, operating mode, and main parameters of the LTF equipment are described in detail. A series of model tests conducted under various conditions yielded over 600 000 high-quality labeled A-scan data. Based on both A-scan and B-scan data types and training strategies, intelligent models with excellent performance for backfill grouting thickness recognition are developed. An intelligent detection method for backfill grouting based on a “cloud-edge-end” architecture is proposed, and the GPR-AI Master platform, based on partial “cloud-edge-end” architecture, is developed to achieve cloud deployment and rapid application of the artificial intelligence model. A dynamic feedback mechanism based on the LTF equipment and intelligent analysis results is proposed, realizing comprehensive monitoring throughout the tunnel excavation process. The application is derived from 16 distinct shield tunnel projects domestically and internationally validate the effectiveness of the intelligent detection method, providing valuable references for the secure and intelligent construction of shield tunnels.