Nodes/ComfyUI-WanAnimalPreprocess/Animal Pose and Detection
ComfyUI Node

Animal Pose and Detection

Detects animal poses from images using ViTPose (AP10k/APT36k) and YOLO. Optionally retargets poses based on a reference image.

By Code2Collapse·Created 7 months ago·Updated 2 months ago· 5
Animal Pose and Detection
  • model
  • images
  • retarget_image
  • pose_data
  • key_frame_body_points
  • bboxes
width832
height480
CategoryWanAnimalPreprocess

Inputs (5)

NameTypeDefaultDescription
modelPOSEMODELFrom ONNX Animal Detection Model Loader.
imagesIMAGEInput video frames as IMAGE batch.
widthINT83264–2048Width of the generation
heightINT48064–2048Height of the generation
retarget_imageoptIMAGEOptional reference image for pose retargeting

Outputs (3)

NameTypeDescription
pose_dataPOSEDATAAnimal pose dict bundle. Feed into Draw Animal ViT Pose.
key_frame_body_pointsSTRINGKey-frame body keypoints as JSON string.
bboxesBBOXPer-frame body BBOX list.