For NVIDIA GPUs, there are numerous GPU-accelerated libraries that are available for different data processing operations across domains.

CV-CUDA is an open-source library that offers GPU-accelerated operators (Python, C++) for image processing/transformation functions like resizing, center crop etc. These are based on hand-optimized kernels that can be faster than some of the other GPU-supported libraries out there. It also has strong interop support with popular DL frameworks like PyTorch and TF with zero-copy interface. https://github.com/CVCUDA/CV-CUDA

Also available for pip install via PyPI: https://pypi.org/project/cvcuda-cu12/ https://pypi.org/project/cvcuda-cu13/