Deploy with Roboflow

GroundingDINO

Grounding DINO is a zero-shot object detection model made by combining a Transformer-based DINO detector and grounded pre-training.

Overview

Grounding DINO is a zero-shot object detection model made by combining a Transformer-based DINO detector and grounded pre-training. Grounding DINO does impressively well in zero-shot object detection, where it achieves an impressive performance on COCO and LVIS without being trained on these datasets directly.

According to the Grounding DINO paper abstract, the model achieves "a 52.5 AP on the COCO detection zero-shot transfer benchmark" as well as SOTA performance on several other important benchmarks.

‍

Overall, GroundingDINO is a robust and flexible model for various object detection tasks, especially those requiring open-set detection capabilities.

Performance

	name	backbone	Data	box AP on COCO	Checkpoint	Config
1	GroundingDINO-T	Swin-T	O365,GoldG,Cap4M	48.4 (zero-shot) / 57.2 (fine-tune)	Github link \| HF link	link

Use This Model

Label Data Automatically with GroundingDINO

You can automatically label a dataset using GroundingDINO with help from Autodistill, an open source package for training computer vision models. You can label a folder of images automatically with only a few lines of code. Below, see our tutorials that demonstrate how to use GroundingDINO to train a computer vision model.

Grounding DINO and YOLO-NAS

Grounding DINO and DETR

Grounding DINO and YOLOv5

Grounding DINO and YOLOv8