443 lines
14 KiB
C#
443 lines
14 KiB
C#
#if !UNITY_WSA_10_0
|
|
|
|
using OpenCVForUnity.CoreModule;
|
|
using OpenCVForUnity.UtilsModule;
|
|
using System;
|
|
using System.Collections.Generic;
|
|
using System.Runtime.InteropServices;
|
|
|
|
namespace OpenCVForUnity.DnnModule
|
|
{
|
|
// C++: class Model
|
|
/**
|
|
* This class is presented high-level API for neural networks.
|
|
*
|
|
* Model allows to set params for preprocessing input image.
|
|
* Model creates net from file with trained weights and config,
|
|
* sets preprocessing input and runs forward pass.
|
|
*/
|
|
|
|
public class Model : DisposableOpenCVObject
|
|
{
|
|
|
|
protected override void Dispose(bool disposing)
|
|
{
|
|
|
|
try
|
|
{
|
|
if (disposing)
|
|
{
|
|
}
|
|
if (IsEnabledDispose)
|
|
{
|
|
if (nativeObj != IntPtr.Zero)
|
|
dnn_Model_delete(nativeObj);
|
|
nativeObj = IntPtr.Zero;
|
|
}
|
|
}
|
|
finally
|
|
{
|
|
base.Dispose(disposing);
|
|
}
|
|
|
|
}
|
|
|
|
protected internal Model(IntPtr addr) : base(addr) { }
|
|
|
|
|
|
public IntPtr getNativeObjAddr() { return nativeObj; }
|
|
|
|
// internal usage only
|
|
public static Model __fromPtr__(IntPtr addr) { return new Model(addr); }
|
|
|
|
//
|
|
// C++: cv::dnn::Model::Model(String model, String config = "")
|
|
//
|
|
|
|
/**
|
|
* Create model from deep learning network represented in one of the supported formats.
|
|
* An order of {code model} and {code config} arguments does not matter.
|
|
* param model Binary file contains trained weights.
|
|
* param config Text file contains network configuration.
|
|
*/
|
|
public Model(string model, string config)
|
|
{
|
|
|
|
|
|
nativeObj = DisposableObject.ThrowIfNullIntPtr(dnn_Model_Model_10(model, config));
|
|
|
|
|
|
}
|
|
|
|
/**
|
|
* Create model from deep learning network represented in one of the supported formats.
|
|
* An order of {code model} and {code config} arguments does not matter.
|
|
* param model Binary file contains trained weights.
|
|
*/
|
|
public Model(string model)
|
|
{
|
|
|
|
|
|
nativeObj = DisposableObject.ThrowIfNullIntPtr(dnn_Model_Model_11(model));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: cv::dnn::Model::Model(Net network)
|
|
//
|
|
|
|
/**
|
|
* Create model from deep learning network.
|
|
* param network Net object.
|
|
*/
|
|
public Model(Net network)
|
|
{
|
|
if (network != null) network.ThrowIfDisposed();
|
|
|
|
nativeObj = DisposableObject.ThrowIfNullIntPtr(dnn_Model_Model_12(network.nativeObj));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: Model cv::dnn::Model::setInputSize(Size size)
|
|
//
|
|
|
|
/**
|
|
* Set input size for frame.
|
|
* param size New input size.
|
|
* <b>Note:</b> If shape of the new blob less than 0, then frame size not change.
|
|
* return automatically generated
|
|
*/
|
|
public Model setInputSize(Size size)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
return new Model(DisposableObject.ThrowIfNullIntPtr(dnn_Model_setInputSize_10(nativeObj, size.width, size.height)));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: Model cv::dnn::Model::setInputSize(int width, int height)
|
|
//
|
|
|
|
/**
|
|
*
|
|
* param width New input width.
|
|
* param height New input height.
|
|
* return automatically generated
|
|
*/
|
|
public Model setInputSize(int width, int height)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
return new Model(DisposableObject.ThrowIfNullIntPtr(dnn_Model_setInputSize_11(nativeObj, width, height)));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: Model cv::dnn::Model::setInputMean(Scalar mean)
|
|
//
|
|
|
|
/**
|
|
* Set mean value for frame.
|
|
* param mean Scalar with mean values which are subtracted from channels.
|
|
* return automatically generated
|
|
*/
|
|
public Model setInputMean(Scalar mean)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
return new Model(DisposableObject.ThrowIfNullIntPtr(dnn_Model_setInputMean_10(nativeObj, mean.val[0], mean.val[1], mean.val[2], mean.val[3])));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: Model cv::dnn::Model::setInputScale(Scalar scale)
|
|
//
|
|
|
|
/**
|
|
* Set scalefactor value for frame.
|
|
* param scale Multiplier for frame values.
|
|
* return automatically generated
|
|
*/
|
|
public Model setInputScale(Scalar scale)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
return new Model(DisposableObject.ThrowIfNullIntPtr(dnn_Model_setInputScale_10(nativeObj, scale.val[0], scale.val[1], scale.val[2], scale.val[3])));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: Model cv::dnn::Model::setInputCrop(bool crop)
|
|
//
|
|
|
|
/**
|
|
* Set flag crop for frame.
|
|
* param crop Flag which indicates whether image will be cropped after resize or not.
|
|
* return automatically generated
|
|
*/
|
|
public Model setInputCrop(bool crop)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
return new Model(DisposableObject.ThrowIfNullIntPtr(dnn_Model_setInputCrop_10(nativeObj, crop)));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: Model cv::dnn::Model::setInputSwapRB(bool swapRB)
|
|
//
|
|
|
|
/**
|
|
* Set flag swapRB for frame.
|
|
* param swapRB Flag which indicates that swap first and last channels.
|
|
* return automatically generated
|
|
*/
|
|
public Model setInputSwapRB(bool swapRB)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
return new Model(DisposableObject.ThrowIfNullIntPtr(dnn_Model_setInputSwapRB_10(nativeObj, swapRB)));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: void cv::dnn::Model::setInputParams(double scale = 1.0, Size size = Size(), Scalar mean = Scalar(), bool swapRB = false, bool crop = false)
|
|
//
|
|
|
|
/**
|
|
* Set preprocessing parameters for frame.
|
|
* param size New input size.
|
|
* param mean Scalar with mean values which are subtracted from channels.
|
|
* param scale Multiplier for frame values.
|
|
* param swapRB Flag which indicates that swap first and last channels.
|
|
* param crop Flag which indicates whether image will be cropped after resize or not.
|
|
* blob(n, c, y, x) = scale * resize( frame(y, x, c) ) - mean(c) )
|
|
*/
|
|
public void setInputParams(double scale, Size size, Scalar mean, bool swapRB, bool crop)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
dnn_Model_setInputParams_10(nativeObj, scale, size.width, size.height, mean.val[0], mean.val[1], mean.val[2], mean.val[3], swapRB, crop);
|
|
|
|
|
|
}
|
|
|
|
/**
|
|
* Set preprocessing parameters for frame.
|
|
* param size New input size.
|
|
* param mean Scalar with mean values which are subtracted from channels.
|
|
* param scale Multiplier for frame values.
|
|
* param swapRB Flag which indicates that swap first and last channels.
|
|
* blob(n, c, y, x) = scale * resize( frame(y, x, c) ) - mean(c) )
|
|
*/
|
|
public void setInputParams(double scale, Size size, Scalar mean, bool swapRB)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
dnn_Model_setInputParams_11(nativeObj, scale, size.width, size.height, mean.val[0], mean.val[1], mean.val[2], mean.val[3], swapRB);
|
|
|
|
|
|
}
|
|
|
|
/**
|
|
* Set preprocessing parameters for frame.
|
|
* param size New input size.
|
|
* param mean Scalar with mean values which are subtracted from channels.
|
|
* param scale Multiplier for frame values.
|
|
* blob(n, c, y, x) = scale * resize( frame(y, x, c) ) - mean(c) )
|
|
*/
|
|
public void setInputParams(double scale, Size size, Scalar mean)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
dnn_Model_setInputParams_12(nativeObj, scale, size.width, size.height, mean.val[0], mean.val[1], mean.val[2], mean.val[3]);
|
|
|
|
|
|
}
|
|
|
|
/**
|
|
* Set preprocessing parameters for frame.
|
|
* param size New input size.
|
|
* param scale Multiplier for frame values.
|
|
* blob(n, c, y, x) = scale * resize( frame(y, x, c) ) - mean(c) )
|
|
*/
|
|
public void setInputParams(double scale, Size size)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
dnn_Model_setInputParams_13(nativeObj, scale, size.width, size.height);
|
|
|
|
|
|
}
|
|
|
|
/**
|
|
* Set preprocessing parameters for frame.
|
|
* param scale Multiplier for frame values.
|
|
* blob(n, c, y, x) = scale * resize( frame(y, x, c) ) - mean(c) )
|
|
*/
|
|
public void setInputParams(double scale)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
dnn_Model_setInputParams_14(nativeObj, scale);
|
|
|
|
|
|
}
|
|
|
|
/**
|
|
* Set preprocessing parameters for frame.
|
|
* blob(n, c, y, x) = scale * resize( frame(y, x, c) ) - mean(c) )
|
|
*/
|
|
public void setInputParams()
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
dnn_Model_setInputParams_15(nativeObj);
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: void cv::dnn::Model::predict(Mat frame, vector_Mat& outs)
|
|
//
|
|
|
|
/**
|
|
* Given the {code input} frame, create input blob, run net and return the output {code blobs}.
|
|
* param outs Allocated output blobs, which will store results of the computation.
|
|
* param frame automatically generated
|
|
*/
|
|
public void predict(Mat frame, List<Mat> outs)
|
|
{
|
|
ThrowIfDisposed();
|
|
if (frame != null) frame.ThrowIfDisposed();
|
|
Mat outs_mat = new Mat();
|
|
dnn_Model_predict_10(nativeObj, frame.nativeObj, outs_mat.nativeObj);
|
|
Converters.Mat_to_vector_Mat(outs_mat, outs);
|
|
outs_mat.release();
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: Model cv::dnn::Model::setPreferableBackend(dnn_Backend backendId)
|
|
//
|
|
|
|
public Model setPreferableBackend(int backendId)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
return new Model(DisposableObject.ThrowIfNullIntPtr(dnn_Model_setPreferableBackend_10(nativeObj, backendId)));
|
|
|
|
|
|
}
|
|
|
|
|
|
//
|
|
// C++: Model cv::dnn::Model::setPreferableTarget(dnn_Target targetId)
|
|
//
|
|
|
|
public Model setPreferableTarget(int targetId)
|
|
{
|
|
ThrowIfDisposed();
|
|
|
|
return new Model(DisposableObject.ThrowIfNullIntPtr(dnn_Model_setPreferableTarget_10(nativeObj, targetId)));
|
|
|
|
|
|
}
|
|
|
|
|
|
#if (UNITY_IOS || UNITY_WEBGL) && !UNITY_EDITOR
|
|
const string LIBNAME = "__Internal";
|
|
#else
|
|
const string LIBNAME = "opencvforunity";
|
|
#endif
|
|
|
|
|
|
|
|
// C++: cv::dnn::Model::Model(String model, String config = "")
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_Model_10(string model, string config);
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_Model_11(string model);
|
|
|
|
// C++: cv::dnn::Model::Model(Net network)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_Model_12(IntPtr network_nativeObj);
|
|
|
|
// C++: Model cv::dnn::Model::setInputSize(Size size)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_setInputSize_10(IntPtr nativeObj, double size_width, double size_height);
|
|
|
|
// C++: Model cv::dnn::Model::setInputSize(int width, int height)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_setInputSize_11(IntPtr nativeObj, int width, int height);
|
|
|
|
// C++: Model cv::dnn::Model::setInputMean(Scalar mean)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_setInputMean_10(IntPtr nativeObj, double mean_val0, double mean_val1, double mean_val2, double mean_val3);
|
|
|
|
// C++: Model cv::dnn::Model::setInputScale(Scalar scale)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_setInputScale_10(IntPtr nativeObj, double scale_val0, double scale_val1, double scale_val2, double scale_val3);
|
|
|
|
// C++: Model cv::dnn::Model::setInputCrop(bool crop)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_setInputCrop_10(IntPtr nativeObj, [MarshalAs(UnmanagedType.U1)] bool crop);
|
|
|
|
// C++: Model cv::dnn::Model::setInputSwapRB(bool swapRB)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_setInputSwapRB_10(IntPtr nativeObj, [MarshalAs(UnmanagedType.U1)] bool swapRB);
|
|
|
|
// C++: void cv::dnn::Model::setInputParams(double scale = 1.0, Size size = Size(), Scalar mean = Scalar(), bool swapRB = false, bool crop = false)
|
|
[DllImport(LIBNAME)]
|
|
private static extern void dnn_Model_setInputParams_10(IntPtr nativeObj, double scale, double size_width, double size_height, double mean_val0, double mean_val1, double mean_val2, double mean_val3, [MarshalAs(UnmanagedType.U1)] bool swapRB, [MarshalAs(UnmanagedType.U1)] bool crop);
|
|
[DllImport(LIBNAME)]
|
|
private static extern void dnn_Model_setInputParams_11(IntPtr nativeObj, double scale, double size_width, double size_height, double mean_val0, double mean_val1, double mean_val2, double mean_val3, [MarshalAs(UnmanagedType.U1)] bool swapRB);
|
|
[DllImport(LIBNAME)]
|
|
private static extern void dnn_Model_setInputParams_12(IntPtr nativeObj, double scale, double size_width, double size_height, double mean_val0, double mean_val1, double mean_val2, double mean_val3);
|
|
[DllImport(LIBNAME)]
|
|
private static extern void dnn_Model_setInputParams_13(IntPtr nativeObj, double scale, double size_width, double size_height);
|
|
[DllImport(LIBNAME)]
|
|
private static extern void dnn_Model_setInputParams_14(IntPtr nativeObj, double scale);
|
|
[DllImport(LIBNAME)]
|
|
private static extern void dnn_Model_setInputParams_15(IntPtr nativeObj);
|
|
|
|
// C++: void cv::dnn::Model::predict(Mat frame, vector_Mat& outs)
|
|
[DllImport(LIBNAME)]
|
|
private static extern void dnn_Model_predict_10(IntPtr nativeObj, IntPtr frame_nativeObj, IntPtr outs_mat_nativeObj);
|
|
|
|
// C++: Model cv::dnn::Model::setPreferableBackend(dnn_Backend backendId)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_setPreferableBackend_10(IntPtr nativeObj, int backendId);
|
|
|
|
// C++: Model cv::dnn::Model::setPreferableTarget(dnn_Target targetId)
|
|
[DllImport(LIBNAME)]
|
|
private static extern IntPtr dnn_Model_setPreferableTarget_10(IntPtr nativeObj, int targetId);
|
|
|
|
// native support for java finalize()
|
|
[DllImport(LIBNAME)]
|
|
private static extern void dnn_Model_delete(IntPtr nativeObj);
|
|
|
|
}
|
|
}
|
|
|
|
#endif |