ComfyUI Node: Segment Face

Segment Face:

The Segment Face node is designed to facilitate the segmentation of facial features from an image using a pre-trained BiSeNet model. This node is particularly useful for AI artists who need to isolate specific parts of a face, such as the skin, eyes, mouth, and optionally the hair and neck, for further processing or manipulation. By leveraging deep learning techniques, the Segment Face node can accurately identify and mask these facial regions, allowing for precise and targeted edits. This capability is essential for tasks such as face swapping, facial feature enhancement, and other creative applications where detailed control over facial components is required.

Segment Face Input Parameters:

face

This parameter expects an image tensor representing the face to be segmented. The image should be in RGB format and will be processed to identify various facial features. The quality and resolution of the input image can significantly impact the accuracy of the segmentation.

model

This parameter requires a pre-trained BiSeNet model, which is used to perform the segmentation. The model should be capable of identifying and classifying different facial regions. The performance of the segmentation largely depends on the quality and training of the provided model.

image

This parameter is an image tensor that represents the entire image containing the face. It is used in conjunction with the segmented face to apply the mask and isolate the desired facial features.

expand

This integer parameter controls the expansion of the mask around the segmented facial features. It helps in ensuring that the mask covers the desired area adequately. The minimum value is 0, and there is no explicit maximum value, but it should be set according to the specific requirements of the task.

include_hair

This parameter is a string that determines whether the hair should be included in the segmentation mask. It accepts values like "enable" to include the hair in the mask. Including hair can be useful for tasks that require the entire head region.

include_neck

This parameter is a string that determines whether the neck should be included in the segmentation mask. It accepts values like "enable" to include the neck in the mask. Including the neck can be beneficial for tasks that need to consider the full head and neck region.

Segment Face Output Parameters:

masked_face

This output is an image tensor representing the face with the applied mask. The masked face image isolates the segmented facial features, making it easier to perform further processing or manipulation on specific parts of the face.

mask

This output is a tensor representing the binary mask of the segmented facial features. The mask highlights the regions of interest, such as the skin, eyes, mouth, and optionally the hair and neck, allowing for precise control over these areas in subsequent operations.

Segment Face Usage Tips:

• Ensure that the input image is of high quality and resolution to achieve better segmentation results.
• Adjust the expand parameter to fine-tune the coverage of the mask around the facial features, especially if you need to include some surrounding areas.
• Use the include_hair and include_neck parameters to customize the segmentation mask according to your specific needs, such as including the entire head region for face swapping tasks.

Segment Face Common Errors and Solutions:

"Invalid model input"

• Explanation: This error occurs when the provided model is not a valid BiSeNet model or is not properly loaded.
• Solution: Ensure that you are using a correctly pre-trained BiSeNet model and that it is properly loaded before passing it to the node.

"Image tensor shape mismatch"

• Explanation: This error happens when the input image tensor does not have the expected shape or format.
• Solution: Verify that the input image tensor is in the correct shape and format (RGB) before passing it to the node. The image should be a 3D tensor with dimensions corresponding to height, width, and channels.

"CUDA out of memory"

• Explanation: This error occurs when the GPU runs out of memory while processing the image.
• Solution: Reduce the size of the input image or use a smaller batch size. Alternatively, try running the process on a machine with more GPU memory.

"Invalid expand parameter"

• Explanation: This error is raised when the expand parameter is set to a negative value or an excessively large value.
• Solution: Ensure that the expand parameter is set to a non-negative integer and is within a reasonable range for your specific task.