GroundingDINO-stk

Welcome to the GroundingDINO-stk project! This repository provides a source that can install the GroundingDINO library from PyPI. The GroundingDINO version was last updated on 2023/06/29.

This document contains all the information you need to get started, including:

• Prerequisites - The basic requirements to use this project.
• Support List - The versions of different libraries and tools that are compatible with this project.
• Installation - Step-by-step instructions on how to install and set up this project.

Prerequisites

The project requires CUDA version 10.2 or later. CUDA is a parallel computing platform and application programming interface (API) model created by Nvidia. It allows software developers and software engineers to use a CUDA-enabled graphics processing unit (GPU) for general-purpose processing.

Support List

Supported Versions

The following table lists the versions of torch, torchvision, and Python that are compatible with this project, along with the respective versions of CUDA they support. Ensure your setup aligns with one of these combinations for smooth operation of the project.

We support the following versions.

• Currently, PyTorch only supported the latest version of Cuda is 11.8.

Unsupported Versions

The following table lists the versions of torch, torchvision, and Python that are not supported by this project. If your setup includes any of these versions, you might encounter issues while using the project.

We didn't support following versions.

Installation

Installation of this project requires careful matching of torch and CUDA versions. The versions you choose should align with the supported versions mentioned above.

Match list

The following table provides a match list of torch and CUDA versions that are known to work together well:

Method 1

This method involves using pip3 to install the package directly. Replace <torch-version> and <cuda-version> with the versions you wish to install. The cuda-version does not include a dot.

# <groundingdino-stk-version> = 1.0.0
# <torch-version> = 2.0
# <cuda-version> = 117
# Example:
pip3 install groundingdino-stk==<groundingdino-stk-version>.<torch-version>.<cuda-version> --extra-index-url https://download.pytorch.org/whl/cu<cuda-version>

Method 2

This method involves modifying and running the install.sh file. Replace the placeholder versions on lines 11 ~ 13 of the install.sh file with the versions of groundingdino-stk, torch and CUDA that you want to install. Then, run the install.sh file.

bash install.sh

Authors/Contributors

• Samir Singh
• Timothius Victorio
• Yichen Kao

Acknowledgements

This project is based on GroundingDINO, adding support for various CUDA, PyTorch, and Python versions compatibility.

https://github.com/IDEA-Research/GroundingDINO

Feel free to contact us if you need further assistance navigating or using this project. We're here to help!