IDEA-Research/GroundingDINO

Health v2 · maintenance only

Adoption not part of the score

10515 stars · 1065 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Official PyTorch implementation of Grounding DINO, an open-set object detector that fuses a Transformer-based DINO detector with grounded vision-language pre-training. It detects arbitrary objects specified by text inputs such as category names or referring expressions, achieving state-of-the-art zero-shot results on COCO, LVIS, ODinW, and RefCOCO benchmarks.

Use cases

• detect arbitrary objects in images using text prompts
• zero-shot object detection without training on target categories
• detect objects by referring expressions like 'the man in a red hat'
• generate bounding boxes to feed into SAM for segmentation
• label datasets automatically with open-vocabulary detection
• integrate open-world detection into robotics or video tracking pipelines

When to choose

• you need to detect objects beyond a fixed set of trained categories using natural language
• you want strong zero-shot detection performance with pretrained checkpoints
• you need a detector backbone to combine with SAM or Grounded SAM pipelines
• you want a Hugging Face-compatible open-vocabulary detection model

When to avoid

• you need a lightweight detector for edge devices with strict latency budgets
• you only need closed-set detection on a fixed category list where a smaller YOLO-style model suffices
• you need a GPU-free CPU-only deployment
• you need a maintained API service rather than a research codebase

Facets

library · maturity stable

computer-vision image-processing machine-learning deep-learning computer-vision machine-learning artificial-intelligence image-processing python cross-platform object-detection open-set-detection vision-language zero-shot-detection transformer pytorch grounded-pretraining eccv-2024 gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="IDEA-Research/GroundingDINO")