---
id: nvidia-nvimgcodec-cu12
version: "0.9.0.20"
license: Apache License 2.0
license_treatment: permissive
maintenance: active
---

# nvidia-nvimgcodec-cu12 — NVIDIA nvimgcodec  for CUDA 12.

License: permissive · Maintenance: active · Downloads: 184.6K/mo

## What it is and what it does

nvImageCodec is NVIDIA's GPU-accelerated image codec library designed as an extensible framework. It provides a unified interface for codec operations, with optional plugins for nvjpeg, nvjpeg2k, and nvtiff formats. The package is built specifically for CUDA 12 and supports Python 3.9 through 3.14.

The library offloads image encoding and decoding to the GPU, reducing CPU overhead for image processing workloads. Installation includes optional dependencies for specific codec support—you can install just the base library or add nvjpeg, nvjpeg2k, nvtiff, or all plugins together. Platform support is limited to x86_64, aarch64, and Windows due to the nature of GPU-specific wheels.

Use it for:

- Accelerate batch image decoding in computer vision pipelines using GPU hardware instead of CPU.
- Encode or decode JPEG images at scale in data preprocessing for machine learning workflows.
- Process TIFF images with GPU acceleration in scientific or medical imaging applications.
- Build custom codec plugins on top of the nvImageCodec framework for specialized image formats.
- Reduce latency in real-time image processing by offloading codec operations to the GPU.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

nvImageCodec provides GPU-accelerated codec operations through a unified interface, with optional plugins for JPEG, JPEG2K, and TIFF encoding and decoding.

Yes, if you have a compatible NVIDIA GPU with CUDA 12 and need GPU-accelerated image codec operations. The library is actively maintained, has no known vulnerabilities, and offers a clean plugin architecture. No, if you lack GPU hardware or require CPU-only image processing.

## Install

pip install nvidia-nvimgcodec-cu12
uv add nvidia-nvimgcodec-cu12
poetry add nvidia-nvimgcodec-cu12

## Installing nvidia-nvimgcodec-cu12

Before you install: Medium install friction due to platform-specific wheels (x86_64, aarch64, Windows) and CUDA 12 dependency. Active maintenance with recent commits and no known vulnerabilities.

License in practice: Apache License 2.0 is permissive, allowing commercial and private use with minimal restrictions.

Quickstart:

pip install nvidia-nvimgcodec-cu12
# For JPEG support:
pip install nvidia-nvimgcodec-cu12[nvjpeg]

import nvidia_nvimgcodec as nvimgcodec
# Use codec operations via the unified interface

Requires CUDA 12 runtime and compatible GPU; wheels available only for x86_64, aarch64, and Windows platforms.

Verify before relying:

- Specific codec operations and API surface beyond the unified interface framework
- Performance benchmarks or throughput claims for accelerated operations
- Compatibility matrix between optional plugins and CUDA versions

## Package facts

- License: Apache License 2.0 (permissive)
- Python support: supports_current
- Install friction: medium
- Maintenance: active
- Downloads: 184.6K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

GPU image codec acceleration, NVIDIA nvimgcodec CUDA 12, accelerated image compression, JPEG GPU encoding decoding, hardware-accelerated image processing, CUDA image codec library, nvjpeg nvtiff acceleration, gpu-acceleration, image-codec, cuda

[View on SkillFed](https://skillfed.io/packages/nvidia-nvimgcodec-cu12) · [View on PyPI](https://pypi.org/project/nvidia-nvimgcodec-cu12/)