Video portrait segmentation algorithm based on twin network

The invention discloses a video portrait segmentation algorithm based on a twin network, and relates to the technical field of image processing, a twin network structure is adopted, and the basic structure of the algorithm comprises a video frame acquisition image module, an RGB separation module, an Encoder network module, an SE module, a Decoder network module and a JPU module; a deep learning PyTorch framework is adopted to construct the modules, a model learning video processing method is adopted, a precise alpha mask is predicted for each frame of a video, a task is extracted from a given image or video, and high-resolution video portrait segmentation in a complex scene is realized. 本发明公开了一种基于孪生网络的视频人像分割算法,涉及图像处理技术领域，采用孪生网络结构，其基本结构包括视频帧获取图像模块、RGB分离模块、Encoder网络模块、SE模块、Decoder网络模块和JPU模块；本发明采用深度学习PyTorch框架构建上述模块，由模型学习视频处理方法，通过对视频每帧预测一个精准的alpha蒙版，从给定的图像或视频中提取任务，实现在复杂场景下的高分辨率视频人像分割。