Abstract: In view of the low detectability and identification rate of the girth weld, crack-like and pinhole defects that generally exist in the oil and gas pipelines, a new generation of ultra-high-definition pipeline magnetic flux leakage (MFL) inline detector was independently developed through theoretical analysis, modeling and simulation, equipment research and development, field application and other links. As for the detector, the probe channel spacing is up to 0.6 mm and the axial sampling spacing is 1 mm, which meet the requirements for storage and collection of massive data. In addition, the acquisition capacity of signal data is increased by 15 times. The MFL detection, deformation detection and positioning detection are integrated into the same detector, so that all kinds of defects can be detected at one time. Meanwhile, the corrosion, deformation and girth weld defects can be aligned. Moreover, the intelligent identification and analysis software based on deep learning was developed and a BP neural network based deep learning model was established, realizing the integrated identification of defects and improving the detectability of pinhole corrosion. Hence, the defects with an area less than 1 t×1 t (where t is the wall thickness of the pipeline) can be identified effectively. In this way, the prominent problem that the pinhole corrosion and girth weld defects cannot be accurately described and quantified with the current three-axis high-definition magnetic flux leakage internal detector is solved preliminarily, which is of great significance for breaking the foreign monopolies and improving the technical level of pipeline integrity in China.