GPU Acceleration for Model Infererence

GPU acceleration is available for model inference to speed up the process. To enable this feature, you will need a GPU with CUDA support and use corresponding version of AnyLabeling.

For PyPi Installation

Install the GPU version of AnyLabeling with the following command:

pip install anylabeling-gpu

For Binary Installation

If you download the binary version of AnyLabeling from the Releases (opens in a new tab), please choose the binary file with GPU in its name.

For Source Installation

If you are using AnyLabeling from source, please set __preferred_device__ to GPU in anylabeling/app_info.py file to enable GPU acceleration. The final code should look like this:

__appname__ = "AnyLabeling"
__appdescription__ = "Effortless data labeling with AI support"
__version__ = "0.2.19"
__preferred_device__ = "GPU"  # GPU or CPU