Image recognition method, device and equipment and storage medium

Abstract

The embodiment of the invention provides an image recognition method and device, equipment and a storage medium, and the method comprises the steps: carrying out the feature extraction of each frame in at least two frames of images, and obtaining the feature data of each frame; performing convolution processing based on time dimension on a part of the feature data of each frame of image to obtaina first processing result; performing convolution processing based on spatial dimension on the other part of the feature data of each frame of image to obtain a second processing result; and determining an identification result corresponding to the at least two frames of images according to the first processing result and the second processing result of each frame of image. Through application ofthe method and the device, the calculation complexity of convolution processing on the feature data can be reduced, and effective space-time modeling on the feature data of the image can be ensured.

Description

technical field [0001] The embodiment of the present application relates to the technical field of artificial intelligence, and relates to but not limited to an image recognition method, device, equipment and storage medium. Background technique [0002] At present, when performing image recognition, the following two methods are usually adopted: the first method adopts the dual-stream method, which takes the red-green-blue (RGB, Red-Green-Blue) image and the optical flow image as input respectively, and through convolution The neural network performs spatio-temporal information modeling, and then performs image classification by fusing dual-stream information; the second method uses spatio-temporal convolution, directly inputs RGB image frames as 3D data into the network, and jointly performs spatio-temporal information extraction and spatio-temporal information fusion . [0003] However, using the dual-stream method requires extracting optical flow data, which is time-con...

AI Technical Summary

Problems solved by technology

[0003] However, using the dual-stream method requires extracting optical flow data, which is time-consuming and labor-intensive, and the calculation of the optical flow branch network is very large; using the method of spatio-temporal convolution, since the spatio-temporal modeling is independent, it will be subject to the model Influence of Computational Complexity and Spatiotemporal Modeling Capabilities

Images