ONNX (Open Neural Network Exchange) is a cross-platform accelerator that works across different frameworks and libraries such as TensorFlow, Keras, PyTorch, and scikit-learn. Microsoft and Facebook developed the product to ease switching between frameworks and libraries. ONNX now has support from AMD, ARM, Qualcomm, IBM, Intel, and others.

Project Background

• Tool: ONNX Runtime
• Author: Microsoft and Facebook
• Initial Release: September 2017
• Type: Machine Learning Accelerator
• License: MIT
• Contains: Set of libraries, executable installers, and a NuGet package manager
• Language Support: C#, C++, C, Python or Java
• Github: microsoft/onnxruntime
• Runs On: Linux, Windows, and MAC
• Hardware: Supports NVIDIA GPUs
• Twitter: onnxruntime
• Stackflow: onnxruntime

Applications

• Regression
• Image recognition
• Number recognition
• Image classification
• Object detection
• Image recognition