🎯 3 Nodes Total - Loader → Detector → Visualizer 🚀 5 Model Families - GroundingDINO, MM-GroundingDINO, OWLv2, Florence-2, YOLO-World 💾 Smart Caching - Instant reload 📦 Batch Processing - Multiple images at once 🎭 Built-in Masks - No separate BboxToMask node needed

Quick Start

cd ComfyUI/custom_nodes/
git clone <repository-url> ComfyUI-Grounding
cd ComfyUI-Grounding
pip install -r requirements.txt

The Nodes

1. Grounding Model Loader

Load any of 10+ models from a single dropdown.

GroundingDINO: SwinT OGC (694MB) - IDEA-Research/grounding-dino-tiny
GroundingDINO: SwinB (938MB) - IDEA-Research/grounding-dino-base
MM-GroundingDINO: Tiny O365+GoldG (50.4 mAP) - openmmlab-community/mm_grounding_dino_tiny_o365v1_goldg
MM-GroundingDINO: Tiny O365+GoldG+GRIT (50.5 mAP) - openmmlab-community/mm_grounding_dino_tiny_o365v1_goldg_grit
MM-GroundingDINO: Tiny O365+GoldG+V3Det (50.6 mAP) - openmmlab-community/mm_grounding_dino_tiny_o365v1_goldg_v3det
MM-GroundingDINO: Base O365+GoldG+V3Det (52.5 mAP) - openmmlab-community/mm_grounding_dino_base_o365v1_goldg_v3det
MM-GroundingDINO: Base All Datasets (59.5 mAP) - openmmlab-community/mm_grounding_dino_base_all
MM-GroundingDINO: Large O365v2+OIv6+GoldG (53.0 mAP) - openmmlab-community/mm_grounding_dino_large_o365v2_oiv6_goldg
MM-GroundingDINO: Large All Datasets (60.3 mAP) - openmmlab-community/mm_grounding_dino_large_all
OWLv2: Base Patch16 - google/owlv2-base-patch16
OWLv2: Large Patch14 - google/owlv2-large-patch14
OWLv2: Base Patch16 Ensemble - google/owlv2-base-patch16-ensemble
OWLv2: Large Patch14 Ensemble - google/owlv2-large-patch14-ensemble ⭐
Florence-2: Base (0.23B params) - microsoft/Florence-2-base ⚡ Best Balance
Florence-2: Large (0.77B params) - microsoft/Florence-2-large ⭐
YOLO-World: v8s (Small) - yolov8s-worldv2.pt ⚡ Fastest
YOLO-World: v8m (Medium) - yolov8m-worldv2.pt
YOLO-World: v8l (Large) - yolov8l-worldv2.pt
YOLO-World: v8x (Extra Large) - yolov8x-worldv2.pt

2. Grounding Detector

Universal detector for all models.

Key Features

Overrides standard text label splitting. It splits only at ".", otherwise label is what you write
Enable single_box_mode for single detection
First load: Downloads model
Second load: Instant from cache
Models stored in ComfyUI-Grounding/models/
Process multiple images in one pass
All nodes support batches

Florence-2 Attention Modes

eager - Most compatible (default)
sdpa - Faster on PyTorch 2.0+
flash_attention_2 - Fastest on A100/H100

2. Bounding Box Visualizer

Re-draw bboxes on images with custom line width. Optional since detector already returns annotated images.

Tips

Prompt Format:

Use periods for multiple labels → "cat . dog ."
Use commas to keep a single label → "cat, dog"

Credits

GroundingDINO - IDEA-Research
OWLv2 - Google Research
Florence-2 - Microsoft Research
YOLO-World - Ultralytics

License

MIT License