groundingdino-py
open-set object detector
What it is and what it does
Grounding DINO combines vision and language understanding to detect objects in images by their natural language descriptions. Unlike traditional object detectors that recognize only pre-trained classes, it can identify any object you describe in text, making it an open-set detector. The model outputs bounding boxes for detected objects along with confidence scores for each word in your text prompt, allowing you to filter results by similarity threshold.
The package is designed for research and production use in computer vision tasks where you need flexible, language-driven object detection. It accepts image-text pairs as input and outputs up to 900 candidate boxes by default, each scored against all input words. The implementation supports both GPU and CPU inference, though GPU is strongly recommended for practical use. Model weights must be downloaded separately from the GitHub releases.
Use it for:
- Automated dataset annotation: use natural language prompts to label objects across large image collections without manual annotation
- Image editing workflows: identify specific objects by description to segment or edit them in combination with tools like Stable Diffusion or SAM
- Visual search and retrieval: find objects matching text descriptions across image databases without retraining for new object types
- Accessibility tools: describe what you want to find in an image and get bounding boxes for screen readers or assistive systems
- Content moderation: detect problematic objects or scenes by describing them in text without maintaining separate classifiers
Worth the install?
AI-flagged interpretation of the facts on this page — verify before relying
Grounding DINO is an open-set object detector that identifies and locates objects in images using natural language descriptions, without requiring training data for specific object classes.
No. While the model itself is capable, the PyPI package is problematic: it has been dormant since December 2023 with no maintenance, installation requires compiling native code with high friction, and the fact sheet lists zero runtime dependencies despite being a deep learning model—suggesting the package metadata is incomplete or the PyPI distribution is not the intended installation path. The GitHub repository is the canonical source; install from there directly if you need this model.
Install
groundingdino-py on PyPI
pip
pip install groundingdino-pyuv
uv add groundingdino-pypoetry
poetry add groundingdino-pyInstalling groundingdino-py
Before you install
Installation requires compiling native code and has high friction. The package is dormant—last commit was 2023-12-20, over a year ago, with no active maintenance. Expect potential compatibility issues with newer Python or dependency versions.
License in practice
Licensed under Apache License 2.0, a permissive license that allows commercial and private use with minimal restrictions, requiring only attribution and notice of modifications.
Quickstart
# Clone and install from source
git clone https://github.com/IDEA-Research/GroundingDINO.git
cd GroundingDINO
pip install -e .
# Download model weights
mkdir weights
cd weights
wget https://github.com/IDEA-Research/GroundingDINO/releases/download/v0.1.0-alpha/groundingdino_swint_ogc.pth
# Basic inference (see demo/inference_on_a_image.py)
CUDA_VISIBLE_DEVICES=0 python demo/inference_on_a_image.py \
-c groundingdino/config/GroundingDINO_SwinT_OGC.py \
-p weights/groundingdino_swint_ogc.pth \
-i image.jpg
Requires CUDA_HOME environment variable to be set for GPU compilation; falls back to CPU-only mode if CUDA is unavailable. Model weights must be downloaded separately. Compilation of native code required during installation.
Verify before relying
- Whether the package works with current PyTorch versions and modern Python releases, given dormant maintenance status since late 2023
- Actual runtime dependencies and their versions, as the fact sheet lists zero runtime dependencies despite being a deep learning model
- Whether the PyPI package is actively maintained or if the GitHub repository is the canonical source
Package facts
| License | Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. "License" shall mean the terms and conditions for… (full text in the JSON record) (permissive) |
| Python support | not specified |
| Install friction | high — source build required |
| Runtime dependencies | none |
| Maintenance | dormant — 1,179 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 133,313/month — #11,520 on PyPI (30-day window, as of 2026-08-14) |
| Known vulnerabilities | none known (OSV.dev, checked 2026-08-14) |
Evidence: groundingdino-py-0.4.0.tar.gz
Tags
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to…
permissive · top 100 on PyPI
huggingface-hubClient library and CLI tool for downloading,…
permissive · top 100 on PyPI
langchainLangChain provides a framework for building…
permissive · top 1,000 on PyPI
hf-xethf-xet provides chunk-based deduplication and…
permissive · top 1,000 on PyPI
tokenizersTokenizers converts raw text into token…
permissive · top 1,000 on PyPI
transformersTransformers provides a unified framework for…
permissive · top 1,000 on PyPI
mmdetMMDetection is an object detection and instance…
permissive · top 15,000 on PyPI
perceptronPerceptron is a Python SDK for building…
unclear · top 15,000 on PyPI
ultralyticsUltralytics YOLO provides a unified framework…
agpl · top 5,000 on PyPI
yolov5YOLOv5 is a packaged object detection model…
copyleft · top 15,000 on PyPI
lightlyLightly provides self-supervised learning…
permissive · top 15,000 on PyPI
rfdetrRF-DETR is a transformer-based object…
permissive · top 15,000 on PyPI
effdetPyTorch implementation of EfficientDet object…
permissive · top 15,000 on PyPI
segmentation-models-pytorchProvides PyTorch-based neural network models…
permissive · top 15,000 on PyPI
speechbrainSpeechBrain is a PyTorch-based toolkit for…
permissive · top 5,000 on PyPI
sahiSAHI performs sliced inference on large images…
permissive · top 15,000 on PyPI