celux-cuda 0.5.1.1

HW accelerated video reading for ML Inference (CUDA version).

CeLux

CeLux is a high-performance Python library for video processing, leveraging the power of FFmpeg. It delivers some of the fastest decode times for full HD videos globally, enabling efficient and seamless video decoding directly into PyTorch tensors.

The name CeLux is derived from the Latin words celer (speed) and lux (light), reflecting its commitment to speed and efficiency.

📚 Documentation

• 📝 Changelog
• 📦 Installation Instructions
  • 🤖 PIP installation
  • 🛠️ Building from Source
• 🚀 Getting Started
• 📊 Benchmarks
• 🤝 Contributing Guide
• ❓ FAQ

🚀 Features

• ⚡ Ultra-Fast Video Decoding: Achieve lightning-fast decode times for full HD videos using hardware acceleration.
• 🔗 Direct Decoding to Tensors: Decode video frames directly into PyTorch tensors for immediate processing.
• 🖥️ Hardware Acceleration Support: Utilize CUDA for GPU-accelerated decoding, significantly improving performance.
• 🔄 Easy Integration: Seamlessly integrates with existing Python workflows, making it easy to incorporate into your projects.

⚡ Quick Start

pip install celux  # cpu only version

pip install celux-cuda  # cuda+cpu

from celux import VideoReader, Scale
#import celux as cx
filters = [Scale(width = 1920, height = 1080)]
reader = VideoReader("/path/to/video.ext",
                    #num_threads: int = os.cpu_count(),
                    device = "cuda",
                    filters = filters
                    )
for frame in reader:
# do something

📊 Benchmark Summary

Library	Device	Frames per Second (FPS)
Celux	CPU	1520.75
Celux	CUDA	1710.85
PyAV	CPU	350.58
OpenCV	CPU	454.44

For more details, see Benchmarks.

📄 License

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). See the LICENSE file for details.

🙏 Acknowledgments

• FFmpeg: The backbone of video processing in CeLux.
• PyTorch: For tensor operations and CUDA support.
• Vcpkg: Simplifies cross-platform dependency management.
• @NevermindNilas: For assistance with testing, API suggestions, and more.