Monitoring-video-stream-scene-orientedNeural network model reasoning acceleration method

The invention discloses a monitoring-video-stream-scene-oriented neural network model reasoning acceleration method, which takes a neural network level as a unit, carries out segmentation according to a neural network reasoning calculation direction, segments an original neural network model into a plurality of modules, and takes the output of the previous module as the input of the next module; parallel processing is carried out on the plurality of divided modules by utilizing a multi-process technology, namely, one module corresponds to one process; the modules of the neural network independently calculate input data in parallel according to a multi-process mode, and data transmission between module processes is carried out in a message queue mode. The invention relates to the technical field of neural network models, and particularly provides a monitoring video stream scene-oriented neural network model reasoning acceleration method, which starts from the feedforward characteristic and hierarchical cascade