45 lines
1.2 KiB
Markdown
45 lines
1.2 KiB
Markdown
# Documentation
|
|
- Class name: FaceToMask
|
|
- Category: ♾️Mixlab/Mask
|
|
- Output node: False
|
|
- Repo Ref: https://github.com/shadowcz007/comfyui-mixlab-nodes.git
|
|
|
|
该节点使用图像处理技术从输入图像中识别和隔离面部特征，专注于绿色通道以创建突出显示检测到的面部的掩码。
|
|
|
|
# Input types
|
|
## Required
|
|
- image
|
|
- 输入图像对于节点的操作至关重要，因为它是提取面部特征和生成掩码的来源。
|
|
- Comfy dtype: IMAGE
|
|
- Python dtype: PIL.Image
|
|
|
|
# Output types
|
|
- MASK
|
|
- 输出掩码代表输入图像中的面部区域，使用绿色通道数据来指示面部的存在。
|
|
- Comfy dtype: TENSOR
|
|
- Python dtype: torch.Tensor
|
|
|
|
# Usage tips
|
|
- Infra type: CPU
|
|
|
|
# Source code
|
|
```
|
|
class FaceToMask:
|
|
|
|
@classmethod
|
|
def INPUT_TYPES(s):
|
|
return {'required': {'image': ('IMAGE',)}}
|
|
RETURN_TYPES = ('MASK',)
|
|
FUNCTION = 'run'
|
|
CATEGORY = '♾️Mixlab/Mask'
|
|
INPUT_IS_LIST = False
|
|
OUTPUT_IS_LIST = (False,)
|
|
|
|
def run(self, image):
|
|
im = tensor2pil(image)
|
|
mask = detect_faces(im)
|
|
mask = pil2tensor(mask)
|
|
channels = ['red', 'green', 'blue', 'alpha']
|
|
mask = mask[:, :, :, channels.index('green')]
|
|
return (mask,)
|
|
``` |