v0.8.2
Summary
Updated the Jetson images so that the default execution provider is CUDA. TensorRT is now an optional configuration via the environment variable ONNXRUNTIME_EXECUTION_PROVIDERS=TensorrtExecutionProvider
. The images are also renamed to:
- roboflow/roboflow-inference-server-jetson-4.5.0
- roboflow/roboflow-inference-server-jetson-4.6.1
- roboflow/roboflow-inference-server-jetson-5.1.1