Abstract: Pipeline transportation is a relatively safe and reliable transportation method for energy media such as oil and natural gas, but pipeline accidents occur from time to time due to manufacturing as well as welding defects, injuries, corrosion, etc., which lay great hidden dangers for the safe operation of pipelines. Magnetic leakage detection is the most widely used non-destructive testing technology in oil and gas pipeline safety inspection, and the analysis of magnetic leakage data is crucial for pipeline safety assessment. In recent years, many scholars have applied deep learning to the data analysis of pipeline leakage detection, and have achieved remarkable results. As a powerful machine learning method, deep learning shows a wide range of application prospects in the field of magnetic leakage detection of oil and gas pipelines. This paper reviews relevant research results at home and abroad, and discussed in detail the research of deep learning in pipeline leakage magnetic detection from three aspects: pipeline target detection, pipeline anomaly classification and pipeline defect quantification. On this basis, it analyzes the future development trend of deep learning in the field of pipeline defect leakage magnetic detection, in order to provide reference for related research.