A fast video inference method based on cyclic residual module

Abstract

The invention provides a fast video inference method based on a cyclic residual module, a cyclic residual module, which reduces the complexity of neural network operation, improves the sparsity of intermediate feature mapping, and synthesizes efficient inference engine, is used in the method. The process is as follows: firstly, frame similarity is utilized in cyclic residual module to reduce theredundant calculation in frame-by-frame video convolutional neural network inference; then, by improving the sparse approximation output of the intermediate feature map, the inference speed is accelerated and the inference precision is ensured by the error control mechanism. Finally, the synthesis of efficient inference engine, that is, the use of matrix vector multiplication in the dynamic sparsity of the accelerator, so that the cycle residual module is in efficient operation. Compared with the traditional method, the invention can remarkably improve the running speed of the vision processing system, and enhance the understanding ability of the network to the real-time changing video fragments, thereby realizing the acceleration of the video inference process on the premise of ensuring the identification accuracy.

Description

technical field [0001] The invention relates to the field of video recognition, in particular to a fast video inference method based on a cyclic residual module. Background technique [0002] With the great breakthroughs in the field of artificial intelligence, the development of video recognition technology has also been greatly promoted; the working principle of video recognition technology is mainly: the front-end video acquisition camera provides clear and stable video signals, and then through the embedded intelligent analysis module in the middle, Identify, detect, and analyze video images, filter out interference, and mark targets and tracks for abnormal situations in video images. In terms of security management, video recognition technology can be used as a monitoring system to effectively monitor and identify people in specific areas, so as to detect illegal and criminal acts in time and track suspects; in human-computer interaction, video recognition technology ca...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the problems of slow inference speed and low precision in the current video inference method, the purpose of the present invention is to provide a fast video inference method based on the cyclic residual module. Firstly, the frame similarity is used in the cyclic residual module to reduce Redundant calculations in frame-by-frame video convolutional neural network inference; then, approximate inference output by improving the sparsity of intermediate feature maps, thereby accelerating the speed of inference, and ensuring the accuracy of inference through an error control mechanism; finally, synthesizing an efficient inference engine , an accelerator that exploits dynamic sparsity in matrix-vector multiplication to make the loop residual module run efficiently

Images