Real-time video portrait segmentation system and method, storage medium and equipment

Abstract

The invention discloses a real-time video portrait segmentation system and method, a storage medium and equipment, and belongs to the technical field of image processing. The method comprises the steps of obtaining a real-time video image, inputting the adjusted video image into a trained convolutional neural network model for feature extraction, and performing video image inference to obtain a portrait segmentation image; wherein the convolutional neural network model adopts an encoding-decoding structure, and the encoding network uses standard convolution, depth separable convolution and a hole convolution layer to extract image features; performing feature fusion reconstruction by the decoding network through a fusion module to complete image feature decoding; and recovering the image by adopting a bilinear difference module, deducing the video image by utilizing a cross entropy loss function, and outputting a segmented portrait segmentation image. According to the method, the adopted convolutional neural network model has few parameters and high operation speed, so that real-time portrait segmentation can be quickly carried out on the video image, and the problem of low segmentation speed of an existing method is solved.

Description

Technical field [0001] The invention belongs to the technical field of image processing, and specifically relates to a system, method, storage medium and equipment for real-time video portrait segmentation. Background technique [0002] Portrait segmentation technology refers to the technology that separates the portrait and background in the image. With the development of technology, portrait segmentation technology is widely used in many fields such as photography, film and television production, video surveillance, etc. Therefore, many portrait segmentation methods have emerged to meet Demands in different fields. In the video field, the requirements for portrait segmentation technology are very high, because compared to static images, video is dynamic and is composed of several consecutive frames of video images. It is necessary to segment portraits from continuously played video images. The technical requirements are very high, so it is very difficult to separate the portra...

AI Technical Summary

Problems solved by technology

[0004] Technical problem: The present invention provides a system, method, storage medium and device for real-time video portrait segmentation, which can quickly and efficiently perform real-time video portrait segmentation using a convolutional neural network model based on depth-separable convolution and atrous convolution. Segmentation, which solves the problem of slow speed when segmenting real-time video portraits

Images