Changelog

v1.7 (Jun 23, 2023)

• Unreal Engine 5.2 version is now available.
• Added DLLEXPORT so that FOnnxTensorInfo can be used from C++ in other modules.
• Added conditional compilation to work with the latest API of ONNX Runtime. See Use latest ONNX Runtime for more details.

v1.6 (Jan 27, 2023)

• Added support for Unreal Engine 5.1.
• Added OnnxModel destructor call when UOnnxModelWrapper is destroyed.

v1.5 (Apr 10, 2022)

• Added support for Unreal Engine 5.0 official release.
  • To bind single-precision float inputs/outputs from Blurprint in UE5, see this page.
  • In UE5, OnnxRuntime module has been renamed to OnnxRuntimeNNEngine module to avoid conflicts with the engine module. Please take care when migrating existing projects.
• Updated OnnxRuntime module
  • Added an option to specify model optimization level.
• Updated demo project
  • The DNN model for facial tracking was changed to one that uses an attention mechanism in Human pose estimation demo.

v1.4 (Mar 18, 2022)

• Updated OnnxRuntime module
  • Added an option to disable dependency to OpenCV.

v1.3 (Mar 04, 2022)

• Updated OnnxRuntime module
  • Added a Blueprint callable function Construct Onnx Tensor Info. Use this function to dynamically specify the shape of input/output tensors.

v1.2 (Feb 18, 2022)

• Updated TextureProcessing module
  • Added a component to convert UTexture to float array. (TextureProcessFloatComponent)
  • Added functions to create UTexture from arrays of byte or float.
  • Fixed a bug that some UTexture cannot be processed by TextureProcessComponent.
    • Now BP_TextureProcessComponent is deprecated. Use TextureProcessComponent instead.
• Updated CustomizedOpenCV module
  • Removed OpenCV's check function to avoid conflict with UE4's check macro.
• Added example projects
  • Added an example for depth estimation using a monocular RGB camera.
  • Added an example for arbitrary artistic style transfer.

v1.1 (Feb 11, 2022)

• Added support for Ubuntu 18.04.6 Desktop 64bit
  • GPU accelerations by CUDA and TensorRT supported.
    • You need an NVIDIA GPU which supports CUDA, cuDNN, and TensorRT.
    • You need to install CUDA ver 11.4.2, cuDNN ver 8.2.4, and TensorRT ver 8.2.3.0.
    • DNN models which contain unsupported operators cannot be loaded when TensorRT is enabled.
      See the official document for supported operators. (NNEngine uses TensorRT 8.2 as backend on Linux)
  • Tested environment:
    • Unreal Engine: 4.26.2, 4.27.2
    • Vulkan utils: 1.1.70+dfsg1-1ubuntu0.18.04.1
    • .NET SDK: 6.0.101-1
    • OS: Ubuntu 18.04.6 Desktop 64bit
    • CPU: Intel i3-8350K
    • GPU: NVIDIA GeForce GTX 1080 Ti
      • Driver: 470.130.01
      • CUDA: 11.4.2-1
      • cuDNN: 8.2.4
      • TensorRT: 8.2.3.0
• Added EXPERIMENTAL support for Android as target build
  • Tested environment:
    • Device: Xiaomi Redmi Note 9S
    • Android version: 10 QKQ1.191215.002
  • Note:
    • You need to convert your model to ORT format.
      See the official document for the details.
    • There are some DNN models which cannot be loaded on Android.
    • NNEngine uses ONNX Runtime Mobile ver 1.8.1 on Android.
  • GPU acceleration by NNAPI is not tested yet.

v1.0 (Dec 21, 2021)

• First release.