Description
Context
Neural networks are graphs consisting of nodes called operators. Each operator corresponds to a mathematical function, usually described in framework's documentation or an AI standard, such as ONNX.
OpenVINO ONNX Frontend is a component responsible for working with ONNX graphs and requires implementation of different ONNX operators in order to use ONNX models. Usually the operator is already implemented in OpenVINO and ONNX Frontend serves as an API translation layer.
This task requires extending OpenVINO ONNX Frontend with DeformConv-19
operator.
Necessary help will be provided by ONNX Fronted team.
What needs to be done?
- Create
.hpp
and.cpp
files forDeformConv
here - Prepare an implementation of this operator in form of a function. It should be placed in opset 1 namespace. You can use the implementation of standard ONNX Pad operator as a reference
- Register the function in ops_bridge.cpp while keeping alphabetical order
- Create test model(s) in ONNX models directory. OpenVINO test infrastructure then converts prototxt files to ONNX models - you will use those models later in tests
- Add tests covering all use cases here
More details on adding operators to ONNX Frontend guide
Note: Since the only difference between DeformConv19
and DeformConv22
is new datatype, maybe ONNX can be expanded with both in the same PR?
Example Pull Requests
- [ONNX] Extend ONNX Frontend with
com.microsoft.Pad
#22000 - ONNX Unique op support #15076
- ONNX CastLike operator tests #14936
- [ONNX FE] Extend ONNX FE for operation GenerateProposals #12510
- More can be found by searching for the phrase "Extend ONNX Frontend" in issues or PRs
Resources
- Contribution guide - start here!
- Intel DevHub Discord channel - engage in discussions, ask questions and talk to OpenVINO developers
- How to link your Pull Request to an issue
- ONNX Operator specification can be found in official ONNX Runtime docs.
- OpenVINO operator specification can be found in OpenVINO docs.
Contact points
Ticket
Metadata
Metadata
Assignees
Type
Projects
Status