Figure dressing method and device based on scene segmentation and computing device

Abstract

The invention discloses a figure dressing method and device based on scene segmentation, and a computing device. The method includes the following steps: a current frame image containing a specific object is obtained in real time; the current frame image is input into a second neural network to perform scene segmentation, and an image of a face part to be dressed of a specific object is obtained;the second neural network is obtained after output data of at least one intermediate layer of the pre-trained first neural network is subjected to guiding and training operation, and the first neuralnetwork is greater than the second neural network in terms of layer quantity. Beautifying effects are exerted on the image of the face part to be dressed in the specific object, fusion with the current frame image is performed, a processed frame image is obtained, and processed video data is obtained after a frame processing image is overlaid on the current frame image; processed video data is displayed. Via use of the figure dressing method and device based on the scene segmentation and the computing device, the neural network with a small number of layers after training operation is used forachieving fast and accurate scene segmentation, the beautifying effects are exerted on the obtained face part to be dressed, and processing effects can be displayed in time.

Description

technical field [0001] The present invention relates to the field of deep learning, in particular to a character dressing method, device, and computing device based on scene segmentation. Background technique [0002] With the development of science and technology, the technology of image acquisition equipment is also improving day by day. The collected images are clearer, the resolution, and the display effect are also greatly improved. However, the existing recorded video is only a monotonous recording material itself, which may not meet the needs of the user. The user wishes to personalize the video. For example, the user wants to add a beautification effect to the facial area of ​​the person in the video. In the prior art, most of the frame images of the video are processed by adaptive blurring, so that the characters in it have a beautifying effect, but the display effect of the video obtained after this processing is not good, and the video is not enough. Sharp, boke...

AI Technical Summary

Problems solved by technology

However, the calculation speed of the multi-layer intermediate layer will be slow, and the segmentation cannot be performed quickly, and the scene segmentation of the image collected by the image acquisition device in real time cannot be performed quickly.

When using a neural network with fewer intermediate layers, the calculation speed is faster due to the fewer intermediate layers, but limited by the number of layers, it may result in limited computing power, poor fitting ability, and inaccurate segmentation results. And other issues

Images