Machine learning-based parking lot parking space state recognition method and system

The invention relates to a machine learning-based parking lot parking space state recognition method and system. The method includes the steps that: the specific installation positions of cameras aredetermined according to the size information of the specific parking spaces of a parking lot; the cameras capture the images of the parking lot in real time, target parking spaces are marked, and marked image data are stored; all vehicles are marked in the stored image data, the marked image data are learned, a parking space map model, a vehicle identification model, and a non-vehicle object modelare established, and feedback is performed continuously, so that the models can be optimized; and the three established models are compared with image data which are acquired in real time, so that arecognition result can be obtained; the recognition result is converted into parking space real-time state information, and the parking space real-time state information is sent to users, and the parking space state information