Beta.2

CV-CUDA 0.4.0 is a major release of the library providing multiple new operators, Jetson Orin support, and updated API documentation.

Release Highlights

CV-CUDA v0.4.0 includes the following key features:

  • 14 new image processing and computer vision operators - Advanced Color Format Conversion - Brightness_Contrast - Color_Twist - FindContours - GaussianNoise - Histogram - Histogram Equalizer - Inpainting - MinAreaRect - MinMaxLoc - Morphology (Open, Close) - On-screen display (Polyline, Point, Line, Text, Rotated Rectangle, Segmented Mask) - RandomResizedCrop - SIFT

  • Updated sample application - Streamed Triton-based Video Segmentation Sample using CV-CUDA and VPF (Video Processing Framework) optimized for performance with video decode/encode on server-side

  • Added Jetson Orin support for core library

  • Updated API documentation

Compatibility

CV-CUDA has been tested on the following compute stack

  • GPU Compute Capability: 7+.x

  • Ubuntu x86_64: 20.04, 22.04

  • CUDA Toolkit: 11.7+ (11.2+ for library build and run)

  • GCC: 11.0+ (9.0 and 10.0 for APIs, with pre-built binary and run)

  • Python: 3.7, 3.8, 3.10

Refer to documentation of the sample applications for dependencies.

Known Issues/Limitations

  • Samples fails for encoding surfaces on T4 with CUDA 11.8 and display driver 520. Suggested workaround is to upgrade to a newer driver 525+.

  • For GCC versions lower than 11.0, C++17 support needs to be enabled when compiling CV-CUDA.

License

CV-CUDA operates under the Apache 2.0 license.

Resources

  1. CV-CUDA GitHub

  2. CV-CUDA Increasing Throughput and Reducing Costs for AI-Based Computer Vision with CV-CUDA

  3. NVIDIA Announces Microsoft, Tencent, Baidu Adopting CV-CUDA for Computer Vision AI

  4. CV-CUDA helps Tencent Cloud audio and video PaaS platform achieve full-process GPU acceleration for video enhancement AI

Acknowledgements

CV-CUDA is developed jointly by NVIDIA and the ByteDance Machine Learning team.