Artificial intelligence CNN and LSTM neural network dynamic identification system

Abstract

The invention discloses an artificial intelligence CNN and LSTM neural network dynamic identification system, which belongs to the field of intelligent security early warning, and comprises a shootingacquisition terminal, a server, a convolutional neural network, a long-short-term memory neural network, an early warning system, a rejection system, a cloud server, a screening system, a confirmation module and a local data module which are connected in sequence, the output end of the elimination system is also connected with the input end of the local data module; the shooting acquisition terminal comprises an intra-area public camera, an intra-area private camera, an intra-area automobile data recorder and an intra-area mobile phone terminal, and authority acquisition APPs are arranged inthe intra-area public camera, the intra-area private camera, the intra-area automobile data recorder and the intra-area mobile phone terminal. The shooting acquisition terminal is connected with the server through the wireless communication unit to realize transmission; according to the system, comprehensive information data acquisition is facilitated, the application range is wide, and the accuracy is high.

Description

technical field [0001] The present invention relates to the field of intelligent security early warning, more specifically, relates to an artificial intelligence CNN, LSTM neural network dynamic identification system. Background technique [0002] Artificial intelligence CNN, LSTM neural network dynamic recognition system is a combination of advanced shooting terminal technology, central processing unit CPU, image processor GPU, neural network processor NPU, heterogeneous / reconfigurable processor technology, convolutional neural network technology, Effective integration of long and short-term memory neural network technology, AI computer early warning processing technology, AI artificial intelligence early warning operation technology, risk factor collection technology, risk factor identification technology, big data analysis technology, cloud computing technology, cloud storage technology, cloud database technology, etc. It is applied to the entire artificial intelligence C...

AI Technical Summary

Problems solved by technology

[0006] (1) The patented shooting terminal The shooting collection terminal (100) is used to collect video streams containing facial features, voice features, and behavioral feature information, and automatically detect and track information on human face, voice, and behavioral features in the image, and then Perform a series of behavior-related technical processing on the detected face features, voice features and behavior feature information, including face recognition, voice recognition, behavior feature information recognition and abnormal behavior recognition (including fighting, theft, falling of the elderly, event, intrusion, etc.), and send the image sequence to the server (200) through wireless communication. The network includes a local area network, the Internet or a wireless network. As we all know, in real life, the shooting terminal (100) can only use Cameras used to serve the public to take pictures and then transmit data are particularly limited in this way: if abnormal behaviors cannot be collected in all directions, if there are no corresponding surveillance cameras in many places, this will cause inability to collect comprehensively Abnormal behavior, how to realize comprehensive information and data collection on the basis of it, this patent cannot be realized, so the limitation is small;

[0007] (2) The convolutional network neural network (300) in this patent is unreasonably designed, which will affect the accuracy of subsequent results

Images