ComfyUI Node
ONNX Animal Detection Model Loader
Loads ONNX models for animal pose detection. Supports both AP10k and APT36k datasets (both use 17 keypoints). Select the dataset matching your ViTPose model.
ONNX Animal Detection Model Loader
- model
◄vitpose_model▾►
◄yolo_model▾►
◄datasetap10k►
◄onnx_deviceCUDAExecutionProvider►
CategoryWanAnimalPreprocess
Inputs (4)
| Name | Type | Default | Description |
|---|---|---|---|
| vitpose_model | COMBO | ViTPose ONNX model for animal pose estimation. Loaded from 'ComfyUI/models/detection' folder. | |
| yolo_model | COMBO | YOLOv8 ONNX model for animal detection. Loaded from 'ComfyUI/models/detection' folder. | |
| dataset | COMBO | ap10k | Dataset the ViTPose model was trained on. AP10k: 23 animal families, good for common animals (cat, dog, horse). APT36k: 30 species, broader coverage and more training data. |
| onnx_device | COMBO | CUDAExecutionProvider | Device to run the ONNX models on |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| model | POSEMODEL | Animal pose model bundle (ViTPose+YOLO+dataset). Connect to `model` on Animal Pose and Detection. |