Skip to content

训练512,推理出的视频嘴部遮罩没有去除。 #195

@Ignorant98

Description

@Ignorant98

我通过data_processing_pipeline.sh将视频处理为512分辨率的,将stage1.yaml配置文件都设置为512,遮罩mask.png也重新输出为512分辨率,并从头训练U-Net,但中途推理出的视频嘴巴一直有个遮罩,在sync confidence值较高时,遮罩相比于低分会淡一些,但我无法确定是否是因为,训练次数和sync confidence较低造成的,希望有大佬能帮我。

val_video_75000.mp4

Image

stage1.txt

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions