Installation fails with CUDA 12.4 and PyTorch 2.6.0 during setup

[winster-bai]

When installing GroundingDINO from source in a conda environment with CUDA 12.4 and PyTorch 2.6.0, the installation fails with errors related to missing pip and torch modules, even though PyTorch is properly installed and CUDA is accessible.

I installed PyTorch using the following command:

pip install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124
PyTorch environment verification shows everything is working correctly

Is CUDA 12.4 officially supported by GroundingDINO?

`Running setup.py develop for groundingdino
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╸━ 30/31 [groundingdino] error: subprocess-exited-with-error

× python setup.py develop did not run successfully.
│ exit code: 1
╰─> [92 lines of output]
    Building wheel groundingdino-0.1.0
    Compiling with CUDA
    running develop
    /home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/setuptools/_distutils/cmd.py:90: DevelopDeprecationWarning: develop command is deprecated.
    !!
    
            ********************************************************************************
            Please avoid running ``setup.py`` and ``develop``.
            Instead, use standards-based tools like pip or uv.
    
            By 2025-Oct-31, you need to update your project and remove deprecated calls
            or your builds will no longer be supported.
    
            See https://github.com/pypa/setuptools/issues/917 for details.
            ********************************************************************************
    
    !!
      self.initialize_options()
    Looking in indexes: https://mirrors.aliyun.com/pypi/simple/
    Obtaining file:///home/dfrobot/GroundingDINO
      Installing build dependencies: started
      Installing build dependencies: finished with status 'done'
      Checking if build backend supports build_editable: started
      Checking if build backend supports build_editable: finished with status 'done'
      Getting requirements to build editable: started
      Getting requirements to build editable: finished with status 'error'
      error: subprocess-exited-with-error
    
      × Getting requirements to build editable did not run successfully.
      │ exit code: 1
      ╰─> [29 lines of output]
          /home/dfrobot/anaconda3/envs/groundingdino/bin/python3.10: No module named pip
          Traceback (most recent call last):
            File "<string>", line 32, in install_torch
          ModuleNotFoundError: No module named 'torch'
    
          During handling of the above exception, another exception occurred:
    
          Traceback (most recent call last):
            File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 389, in <module>
              main()
            File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 373, in main
              json_out["return_val"] = hook(**hook_input["kwargs"])
            File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 157, in get_requires_for_build_editable
              return hook(config_settings)
            File "/tmp/pip-build-env-nd48_00q/overlay/lib/python3.10/site-packages/setuptools/build_meta.py", line 473, in get_requires_for_build_editable
              return self.get_requires_for_build_wheel(config_settings)
            File "/tmp/pip-build-env-nd48_00q/overlay/lib/python3.10/site-packages/setuptools/build_meta.py", line 331, in get_requires_for_build_wheel
              return self._get_build_requires(config_settings, requirements=[])
            File "/tmp/pip-build-env-nd48_00q/overlay/lib/python3.10/site-packages/setuptools/build_meta.py", line 301, in _get_build_requires
              self.run_setup()
            File "/tmp/pip-build-env-nd48_00q/overlay/lib/python3.10/site-packages/setuptools/build_meta.py", line 512, in run_setup
              super().run_setup(setup_script=setup_script)
            File "/tmp/pip-build-env-nd48_00q/overlay/lib/python3.10/site-packages/setuptools/build_meta.py", line 317, in run_setup
              exec(code, locals())
            File "<string>", line 37, in <module>
            File "<string>", line 34, in install_torch
            File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/subprocess.py", line 369, in check_call
              raise CalledProcessError(retcode, cmd)
          subprocess.CalledProcessError: Command '['/home/dfrobot/anaconda3/envs/groundingdino/bin/python3.10', '-m', 'pip', 'install', 'torch']' returned non-zero exit status 1.
          [end of output]
    
      note: This error originates from a subprocess, and is likely not a problem with pip.
    error: subprocess-exited-with-error
    
    × Getting requirements to build editable did not run successfully.
    │ exit code: 1
    ╰─> See above for output.
    
    note: This error originates from a subprocess, and is likely not a problem with pip.
    Traceback (most recent call last):
      File "<string>", line 2, in <module>
      File "<pip-setuptools-caller>", line 35, in <module>
      File "/home/dfrobot/GroundingDINO/setup.py", line 204, in <module>
        setup(
      File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/setuptools/__init__.py", line 115, in setup
        return distutils.core.setup(**attrs)
      File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/setuptools/_distutils/core.py", line 186, in setup
        return run_commands(dist)
      File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/setuptools/_distutils/core.py", line 202, in run_commands
        dist.run_commands()
      File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/setuptools/_distutils/dist.py", line 1002, in run_commands
        self.run_command(cmd)
      File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/setuptools/dist.py", line 1102, in run_command
        super().run_command(command)
      File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/setuptools/_distutils/dist.py", line 1021, in run_command
        cmd_obj.run()
      File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/site-packages/setuptools/command/develop.py", line 39, in run
        subprocess.check_call(cmd)
      File "/home/dfrobot/anaconda3/envs/groundingdino/lib/python3.10/subprocess.py", line 369, in check_call
        raise CalledProcessError(retcode, cmd)
    subprocess.CalledProcessError: Command '['/home/dfrobot/anaconda3/envs/groundingdino/bin/python3.10', '-m', 'pip', 'install', '-e', '.', '--use-pep517', '--no-deps']' returned non-zero exit status 1.
    [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.

error: subprocess-exited-with-error
`
The error seems to occur during the "Getting requirements to build editable" phase, where the setup script attempts to run a subprocess that fails to recognize the existing torch installation.

[winster-bai]

update: i installed cuda-12.1,stll the same error

[Stella-xixi]

请问您的错误解决了吗？

[rachung2510]

Seems to be due to the latest version of pip 25.3, released 6 days ago. This worked for me

pip install "pip<25.3"
pip install --no-cache-dir -e GroundingDINO

[negativegluon]

I encountered the same issue when installing using CUDA 12.2 and PyTorch 2.9.0. The error is the same. However, when I downgrade pip into 25.2(or even 24), the same installation errors happened again.

[HusterRC]

I met the same problem, someone solved it?

[Clamours]

same problem, but I created a new python==3.8 conda env to install and solved it.

[Angusliuuu]

same problem, but I created a new python==3.8 conda env to install and solved it.

yeah, it worked

[ghostcipher1]

@HusterRC , @Clamours , @Abby-simla , @rachung2510 , @winster-bai

Hi everyone,

I wanted to share an update for those following GroundingDINO development. Because the original repository hasn’t received updates in quite some time, I’ve created and published an up-to-date, fully compatible fork that brings the project current with modern tool-chains.

📦 New release: groundingdino-cu128 on PyPI

• Supports PyTorch 2.7 and CUDA 12.8
• Modernized C++/CUDA extension (migrated from TH/THC → ATen/Torch)
• C++17 build system, cleaner setup for Linux and containerized environments
• Ready-to-Run Docker image w\ PyTorch & CUDA support built-in
• minimal tests included for reproducibility
• Learn more, star and fork my repo at GhostCipher

All original license terms (Apache 2.0) and attributions have been preserved in full.
The fork’s changes are limited to modernization, packaging, and build maintenance—no original intellectual property or authorship has been altered or removed.

[fyw1999]

I also encountered this problem and successfully solved it with the help of GPT.
Here’s how you can fix it step by step:

1. Copy the pyproject.toml file from project groundingdino-cu128 to this project.

(or simply use the groundingdino-cu128 repo directly — thanks to the author!)

2. Delete all the requirements related to torch in both pyproject.toml and requirements.txt.

(⚠️ Make sure PyTorch is already installed in your current environment before continuing!)

3. Replace

sources = [os.path.join(extensions_dir, s) for s in sources]
include_dirs = [extensions_dir]

with

sources = [os.path.relpath(s, start=os.path.dirname(__file__)) for s in sources]
include_dirs = [os.path.relpath(extensions_dir, start=os.path.dirname(__file__))]

in setup.py.
4. Reinstall the package:
pip install -e . --no-build-isolation

If it not works, you can directly use project groundingdino-cu128 and try step 2-4, it works for me. My python is 3.11 and pytorch is 2.5.0(py3.11_cuda12.4_cudnn9.1.0_0).

[ghostcipher1]

@fyw1999 ,

Thanks for confirming this workaround — and I appreciate you mentioning groundingdino-cu128. I will be adding additional updates in the days to come.

-GhostCipher

[fyw1999]

@fyw1999 ,

Thanks for confirming this workaround — and I appreciate you mentioning groundingdino-cu128. I will be adding additional updates in the days to come.

-GhostCipher

Thanks a lot for your effort! You really saved me from a huge headache.

[shantamsrivastava]

I also encountered this problem and successfully solved it with the help of GPT. Here’s how you can fix it step by step:

1. Copy the pyproject.toml file from project groundingdino-cu128 to this project.

(or simply use the groundingdino-cu128 repo directly — thanks to the author!)

2. Delete all the requirements related to torch in both pyproject.toml and requirements.txt.

(⚠️ Make sure PyTorch is already installed in your current environment before continuing!)

3. Replace

sources = [os.path.join(extensions_dir, s) for s in sources]
include_dirs = [extensions_dir]

with

sources = [os.path.relpath(s, start=os.path.dirname(__file__)) for s in sources]
include_dirs = [os.path.relpath(extensions_dir, start=os.path.dirname(__file__))]

in setup.py. 4. Reinstall the package: pip install -e . --no-build-isolation

If it not works, you can directly use project groundingdino-cu128 and try step 2-4, it works for me. My python is 3.11 and pytorch is 2.5.0(py3.11_cuda12.4_cudnn9.1.0_0).

This worked for me. Thanks!

[ghostcipher1]

@fyw1999 , @shantamsrivastava,

I wanted to let you know I archived the project. I’ve been refactoring the original works and have created a new, more modern version in a new repo GroundedDINO-VL. If you have any questions etc. please DME or post to the new repo.

GhostCipher

[Joycexqn]

@fyw1999 , @shantamsrivastava,

I wanted to let you know I archived the project. I’ve been refactoring the original works and have created a new, more modern version in a new repo GroundedDINO-VL. If you have any questions etc. please DME or post to the new repo.

GhostCipher

Excuse me,is the "pyproject.toml" in the new repo the same as the one in "[groundingdino-cu128]"(which you have deleted)? cause when i copy the new file as others did,i encounter an "error: no suitable conversion function from "const at::DeprecatedTypeProperties" to "c10::ScalarType" exists".(my env is PyTorch: 2.8.0+cu128

[xht252]

pip install -e . --no-build-isolationis good for me
torch 2.6.0 、torchaudio 2.6.0、torchvision 0.21.0、Python 3.10.19、CUDA Version: 12.2