AI Accelerator Interfaces (AI²)
AI² extend HALCON with support for hardware-accelerated inference of deep learning models. They provide a common programming interface for different hardware platforms.
The AI² currently include the ONNX Runtime AI² Interface, the OpenVINO AI² Interface, and the TensorRT AI² Interface.
Note
AI² are distributed separately from HALCON and can be installed using the Software Manager (SOM).
-
Run deep learning models through ONNX Runtime using execution providers optimized for different hardware platforms.
Execution Providers
-
OpenVINO AI² Interface
Accelerate inference on Intel hardware using the OpenVINO AI² Interface.
See the OpenVINO AI² Interface documentation in the HALCON documentation for details.
-
TensorRT AI² Interface
Accelerate inference on NVIDIA GPUs using the TensorRT AI² Interface.
See the TensorRT AI² Interface documentation in the HALCON documentation for details.