Transform estimator python bindings

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Orin
1

Hi,

Is there python bindings for the Transform estimator algorithm of the VPI library? (I have not found them in the Python API reference)
Is this something planned or do I need to switch to C?

Thank you very much

3

Hi,

Yes, you can find it in the below link:

/opt/nvidia/vpi3/src/python/TransformEstimator.cpp

...
    m.def("transform_estimator", &TransformEstimator, "source"_a, "target"_a, py::kw_only(), "backend"_a = BackendSet{},
          "matches"_a = nullptr, "out_xform"_a = nullptr, "out_inliers"_a = nullptr, "range_rotation"_a = std::nullopt,
          "range_xscale"_a = std::nullopt, "range_yscale"_a = std::nullopt, "range_xtranslation"_a = std::nullopt,
          "range_ytranslation"_a = std::nullopt, "range_shear"_a = std::nullopt,
          "is_affine"_a        = params.xfcfg.constrainedHomography2D.isAffine,
          "max_solver_iters"_a = params.solverMaxIterations, "max_ransac_iters"_a = params.ransacMaxIterations,
          "ransac_reproj_error_tolerance"_a = params.ransacReprojErrorTolerance,
          "ransac_confidence_level"_a       = params.ransacConfidenceLevel,
          "max_refinement_iters"_a = params.maxRefinementIterations, "stream"_a = nullptr, R"pbdoc(

        Estimate transform that takes keypoints from source to target.

        See also:
            Refer to the :vpic:`algorithm explanation<algo_xform_estim>`
            for more details and usage examples.

        Args:
            source (vpi.Array): Array with source keypoints.
            target (vpi.Array): Array with target keypoints.
            backend (vpi.Backend, optional): The backend to be used by the algorithm.
            matches (vpi.Array, optional): Array with correspondence between source and target keypoints.
            stream (vpi.Stream, optional): The stream to be used by the algorithm.
            out_xform (vpi.Array, optional): Array that will store the estimated transform. If not specified, the function will return a new array.
            out_inliers (vpi.Array, optional): Array with the inlier matches.
            range_rotation (Tuple[float,float], optional): range with minimum and maximum rotation.
            range_xscale (Tuple[float,float], optional): range with minimum and maximum horizontal scaling.
           ...

Thanks.