The Polarity Inversion Line (PIL) of solar magnetic field is one of the most important features on the solar surface for studying solar activities, analyzing the field structural evolution, and predicting solar flares. The PILs are the locations where solar flares and dark filaments may occur. The Advanced Space-based Solar Astronomical Observatory (ASO-S) is the first space satellite for solar observations in China. The Full-Disc Vector Magnetograph (FMG) onboard it is used mainly to detect full-disk solar vector magnetic fields with both high spatial and temporal resolutions. In order to improve the efficiency of data processing and better monitor solar activities, we develop one automatic image recognition and processing technology to detect the PILs. It bases on the support vector machine (SVM) model that converts the PIL detection into a binary classification problem in pattern recognition. The algorithm was tested on the vector magnetograms of the Helioseismic and Magnetic Imager (HMI) onboard Solar Dynamics Observatory (SDO). Compared with a more commonly used algorithm, our results show that it can detect the PIL of active regions more accurately and intuitively.