The ammonia chromogenic reaction method is applied to the leakage detection of Liquefied Natural Gas (LNG) containment systems and the effectiveness of the YOLOv7 target detection model in detecting leakage points in ammonia chromogenic images is explored. Addressing the leakage detection needs of LNG membrane containment systems, the principle of the ammonia chromogenic reaction is analyzed. Based on this principle, the necessary components for leakage detection, including the gas injection system, chromogenic spray, and image acquisition terminal, are constructed. For ammonia chromogenic leakage images, the blue leakage points are annotated and the data are enhanced using MixUp data augmentation. For leakage point identification, the YOLOv7 target detection network is trained, achieving mean Average Precision (mAP) of 97.71% on this dataset and meeting the automated auxiliary detection requirements for LNG containment systems.