Video behavior identification method and system based on space enhancement module

Abstract

The invention discloses a video behavior recognition method and system based on a space enhancement module, and the method comprises the following steps: decoding a to-be-detected video into a frame sequence, and storing the decoded frame sequence in an image form; dividing a video into a plurality of video clips by adopting a sparse sampling strategy, extracting a frame from each video clip, and combining to form a stacked frame sequence; calculating a mean value of three channels of all training video frames in the behavior recognition data set, and subtracting the calculated mean value from the sampled frame image; using a residual neural network 3D-ResNet-18 as a backbone network, and enabling a spatial enhancement module to construct a behavior recognition classification network; setting training parameters, inputting the training set into a behavior recognition classification network for training, and storing trained network parameters; in the model deployment stage, fusing linear operators of the space enhancement module; and inputting a to-be-detected video into the behavior recognition classification network, and outputting a final classification result. According to the invention, the behavior recognition effect is improved, and both effectiveness and universality are achieved.

Description

technical field [0001] The invention relates to the technical field of video behavior recognition of computer vision, in particular to a method and system for video behavior recognition based on a spatial enhancement module. Background technique [0002] Human behavior recognition has always been a key research issue in the field of computer vision. Through the study of human body posture and behavior, abnormal behaviors in public places can be detected, such as: running fast, falling, hitting people, etc. Applying deep learning theory to computer vision applications can design high-precision and high-efficiency behavior recognition algorithms. A highly practical behavior recognition algorithm has played an important role in guaranteeing public safety, and it also enables the computer to automatically complete the detection of abnormal human behaviors, saving the time for managers to search manually. The purpose of video behavior recognition is to identify the specific cat...

AI Technical Summary

Problems solved by technology

However, its joint learning of spatio-temporal features leads to a significant attenuation in the ability to extract spatial features alone compared with 2D convolution, so the classification accuracy on data sets with strong spatial features and weak timing is weaker than that of 2D convolution.

Images