add ONNX OP sign, shrink and reciprocal

Mateusdang · Apr 7, 2022 · e36948c · e36948c
1 parent 9aa6470
commit e36948c
Show file tree

Hide file tree

Showing 10 changed files with 284 additions and 7 deletions.
diff --git a/modules/dnn/include/opencv2/dnn/all_layers.hpp b/modules/dnn/include/opencv2/dnn/all_layers.hpp
@@ -794,6 +794,26 @@ CV__DNN_INLINE_NS_BEGIN
         static Ptr<ActivationLayerInt8> create(const LayerParams &params);
     };
 
+    class CV_EXPORTS SignLayer : public ActivationLayer
+    {
+    public:
+        static Ptr<SignLayer> create(const LayerParams &params);
+    };
+
+    class CV_EXPORTS ShrinkLayer : public ActivationLayer
+    {
+    public:
+        float bias;
+        float lambd;
+        static Ptr<ShrinkLayer> create(const LayerParams &params);
+    };
+
+    class CV_EXPORTS ReciprocalLayer : public ActivationLayer
+    {
+    public:
+        static Ptr<ReciprocalLayer> create(const LayerParams &params);
+    };
+
     /* Layers used in semantic segmentation */
 
     class CV_EXPORTS CropLayer : public Layer

diff --git a/modules/dnn/src/cuda/activations.cu b/modules/dnn/src/cuda/activations.cu
@@ -248,6 +248,21 @@ void selu(const Stream& stream, Span<T> output, View<T> input, T alpha, T gamma)
     generic_op<T, SeluFunctor<T>>(stream, output, input, {alpha, gamma});
 }
 
+template <class T>
+void sign(const Stream& stream, Span<T> output, View<T> input) {
+    generic_op<T, SignFunctor<T>>(stream, output, input);
+}
+
+template <class T>
+void shrink(const Stream& stream, Span<T> output, View<T> input, T bias, T lambd) {
+    generic_op<T, ShrinkFunctor<T>>(stream, output, input, {bias, lambd});
+}
+
+template <class T>
+void reciprocal(const Stream& stream, Span<T> output, View<T> input) {
+    generic_op<T, SignFunctor<T>>(stream, output, input);
+}
+
 template <class T>
 void thresholdedrelu(const Stream& stream, Span<T> output, View<T> input, T alpha) {
     generic_op<T, ThresholdedReluFunctor<T>>(stream, output, input, {alpha});
@@ -312,6 +327,9 @@ template void selu<__half>(const Stream&, Span<__half>, View<__half>, __half, __
 template void thresholdedrelu<__half>(const Stream&, Span<__half>, View<__half>, __half);
 template void power<__half>(const Stream&, Span<__half>, View<__half>, __half, __half, __half);
 template void exp<__half>(const Stream&, Span<__half>, View<__half>, __half, __half);
+template void sign<__half>(const Stream&, Span<__half>, View<__half>);
+template void shrink<__half>(const Stream&, Span<__half>, View<__half>, __half, __half);
+template void reciprocal<__half>(const Stream&, Span<__half>, View<__half>);
 #endif
 
 
@@ -351,6 +369,9 @@ template void selu<float>(const Stream&, Span<float>, View<float>, float, float)
 template void thresholdedrelu<float>(const Stream&, Span<float>, View<float>, float);
 template void power<float>(const Stream&, Span<float>, View<float>, float, float, float);
 template void exp<float>(const Stream&, Span<float>, View<float>, float, float);
+template void sign<float>(const Stream&, Span<float>, View<float>);
+template void shrink<float>(const Stream&, Span<float>, View<float>, float, float);
+template void reciprocal<float>(const Stream&, Span<float>, View<float>);
 
 template <class T, std::size_t N> static
 void launch_vectorized_axiswise_relu(const Stream& stream, Span<T> output, View<T> input, std::size_t inner_size, View<T> slope) {

diff --git a/modules/dnn/src/cuda/functors.hpp b/modules/dnn/src/cuda/functors.hpp
@@ -726,6 +726,50 @@ struct DivFunctor {
     CUDA4DNN_DEVICE T operator()(T x, T y) { return x / y; }
 };
 
+template <class T>
+struct SignFunctor {
+    struct Params {
+        CUDA4DNN_HOST_DEVICE Params() {}
+    };
+
+    CUDA4DNN_DEVICE SignFunctor() : SignFunctor(Params{}) { }
+
+    CUDA4DNN_DEVICE T operator()(T value) {
+        return value > T(0) ? T(1) : (value < T(0) ? T(-1) : T(0));
+    }
+};
+
+template <class T>
+struct ShrinkFunctor {
+    struct Params {
+        CUDA4DNN_HOST_DEVICE Params() : bias(0), lambd(0.5) { }
+        CUDA4DNN_HOST_DEVICE Params(T bias_, T lambd_) : bias(bias_), lambd(lambd_) { }
+        T bias, lambd;
+    };
+
+    CUDA4DNN_DEVICE ShrinkFunctor() : bias(0), lambd(0.5) { }
+    CUDA4DNN_DEVICE ShrinkFunctor(const Params& params) : bias{params.bias}, lambd{params.lambd} { }
+
+    CUDA4DNN_DEVICE T operator()(T value) {
+        return value > lambd ? value - bias : (value < -lambd ? value + bias : T(0));
+    }
+
+    T bias, lambd;
+};
+
+template <class T>
+struct ReciprocalFunctor {
+    struct Params {
+        CUDA4DNN_HOST_DEVICE Params() {}
+    };
+
+    CUDA4DNN_DEVICE ReciprocalFunctor() : ReciprocalFunctor(Params{}) { }
+
+    CUDA4DNN_DEVICE T operator()(T value) {
+        return T(1.0f)/value;
+    }
+};
+
 }}}} /* namespace cv::dnn::cuda4dnn::kernels */
 
 #endif /* OPENCV_DNN_SRC_CUDA_FUNCTORS_HPP */
diff --git a/modules/dnn/src/cuda4dnn/kernels/activations.hpp b/modules/dnn/src/cuda4dnn/kernels/activations.hpp
@@ -123,6 +123,14 @@ namespace cv { namespace dnn { namespace cuda4dnn { namespace kernels {
     template <class T>
     void exp(const csl::Stream& stream, csl::Span<T> output, csl::View<T> input, T normScale, T normShift);
 
+    template <class T>
+    void sign(const csl::Stream& stream, csl::Span<T> output, csl::View<T> input);
+
+    template <class T>
+    void shrink(const csl::Stream& stream, csl::Span<T> output, csl::View<T> input, T bias, T lambd);
+
+    template <class T>
+    void reciprocal(const csl::Stream& stream, csl::Span<T> output, csl::View<T> input);
 }}}} /* namespace cv::dnn::cuda4dnn::kernels */
 
 #endif /* OPENCV_DNN_SRC_CUDA4DNN_KERNELS_ACTIVATIONS_HPP */
diff --git a/modules/dnn/src/cuda4dnn/primitives/activation.hpp b/modules/dnn/src/cuda4dnn/primitives/activation.hpp
@@ -584,6 +584,52 @@ namespace cv { namespace dnn { namespace cuda4dnn {
         const T normScale, normShift;
     };
 
+    template <class T>
+    class ShrinkOp final : public BaseOp<ShrinkOp, T> {
+    public:
+        ShrinkOp(csl::Stream stream_, T bias_, T lambd_)
+                : stream(std::move(stream_)), bias{ bias_ }, lambd{ lambd_ } { }
+
+        void calculate(csl::TensorSpan<T> output, csl::TensorView<T> input) const
+        {
+            kernels::shrink<T>(stream, output, input, bias, lambd);
+        }
+
+    private:
+        csl::Stream stream;
+        const T bias, lambd;
+    };
+
+    template <class T>
+    class SignOp final : public BaseOp<SignOp, T> {
+    public:
+        SignOp(csl::Stream stream_)
+                : stream(std::move(stream_)) { }
+
+        void calculate(csl::TensorSpan<T> output, csl::TensorView<T> input) const
+        {
+            kernels::sign<T>(stream, output, input);
+        }
+
+    private:
+        csl::Stream stream;
+    };
+
+    template <class T>
+    class ReciprocalOp final : public BaseOp<ReciprocalOp, T> {
+    public:
+        ReciprocalOp(csl::Stream stream_)
+                : stream(std::move(stream_)) { }
+
+        void calculate(csl::TensorSpan<T> output, csl::TensorView<T> input) const
+        {
+            kernels::reciprocal<T>(stream, output, input);
+        }
+
+    private:
+        csl::Stream stream;
+    };
+
 }}} /* namespace cv::dnn::cuda4dnn */
 
 #endif /* OPENCV_DNN_SRC_CUDA4DNN_PRIMITIVES_ACTIVATION_HPP */
diff --git a/modules/dnn/src/init.cpp b/modules/dnn/src/init.cpp
@@ -130,6 +130,8 @@ void initializeLayerFactory()
     CV_DNN_REGISTER_LAYER_CLASS(HardSwish,      HardSwishLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Sin,            SinLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Sinh,           SinhLayer);
+    CV_DNN_REGISTER_LAYER_CLASS(Sign,           SignLayer);
+    CV_DNN_REGISTER_LAYER_CLASS(Shrink,         ShrinkLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Softplus,       SoftplusLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Softsign,       SoftsignLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Tan,            TanLayer);
@@ -144,6 +146,7 @@ void initializeLayerFactory()
     CV_DNN_REGISTER_LAYER_CLASS(Silence,        BlankLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Const,          ConstLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Arg,            ArgLayer);
+    CV_DNN_REGISTER_LAYER_CLASS(Reciprocal,     ReciprocalLayer);
 
     CV_DNN_REGISTER_LAYER_CLASS(Crop,           CropLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Eltwise,        EltwiseLayer);

diff --git a/modules/dnn/src/layers/elementwise_layers.cpp b/modules/dnn/src/layers/elementwise_layers.cpp
@@ -2270,6 +2270,96 @@ struct ChannelsPReLUFunctor : public BaseFunctor
     int64 getFLOPSPerElement() const { return 1; }
 };
 
+struct SignFunctor : public BaseDefaultFunctor<SignFunctor>
+{
+    typedef SignLayer Layer;
+
+    bool supportBackend(int backendId, int)
+    {
+        return backendId == DNN_BACKEND_OPENCV ||
+               backendId == DNN_BACKEND_CUDA;
+    }
+
+    inline float calculate(float x) const
+    {
+        return x > 0 ? 1 : (x < 0 ? -1 : 0);
+    }
+
+#ifdef HAVE_CUDA
+    Ptr<BackendNode> initCUDA(int target, csl::Stream stream)
+    {
+        return make_cuda_node<cuda4dnn::SignOp>(target, stream);
+    }
+#endif
+
+    int64 getFLOPSPerElement() const { return 1; }
+};
+
+template<>
+const char* const SignFunctor::BaseDefaultFunctor<SignFunctor>::ocl_kernel_name = "SignForward";
+
+
+struct ShrinkFunctor : public BaseDefaultFunctor<ShrinkFunctor>
+{
+    typedef ShrinkLayer Layer;
+    float bias;
+    float lambd;
+
+    explicit ShrinkFunctor(float bias_ = 0.0f, float lambd_ = 0.5f) : bias(bias_), lambd(lambd_) {}
+
+    bool supportBackend(int backendId, int)
+    {
+        return backendId == DNN_BACKEND_OPENCV ||
+               backendId == DNN_BACKEND_CUDA;
+    }
+
+    inline float calculate(float x) const
+    {
+        return x > lambd ? x - bias : (x < -lambd ? x + bias : 0);
+    }
+
+#ifdef HAVE_CUDA
+    Ptr<BackendNode> initCUDA(int target, csl::Stream stream)
+    {
+        return make_cuda_node<cuda4dnn::ShrinkOp>(target, stream);
+    }
+#endif
+
+    int64 getFLOPSPerElement() const { return 1; }
+};
+
+template<>
+const char* const ShrinkFunctor::BaseDefaultFunctor<ShrinkFunctor>::ocl_kernel_name = "ShrinkForward";
+
+struct ReciprocalFunctor : public BaseDefaultFunctor<ReciprocalFunctor>
+{
+    typedef ReciprocalLayer Layer;
+
+    bool supportBackend(int backendId, int)
+    {
+        return backendId == DNN_BACKEND_OPENCV ||
+               backendId == DNN_BACKEND_CUDA;
+    }
+
+    inline float calculate(float x) const
+    {
+        return 1.0/x;
+    }
+
+#ifdef HAVE_CUDA
+    Ptr<BackendNode> initCUDA(int target, csl::Stream stream)
+    {
+        return make_cuda_node<cuda4dnn::ReciprocalOp>(target, stream);
+    }
+#endif
+
+    int64 getFLOPSPerElement() const { return 1; }
+};
+
+template<>
+const char* const ReciprocalFunctor::BaseDefaultFunctor<ReciprocalFunctor>::ocl_kernel_name = "ReciprocalForward";
+
+
 #define ACTIVATION_CREATOR_FOR(_Layer, _Functor, ...) \
 Ptr<_Layer> _Layer::create() { \
     return return Ptr<_Layer>( new ElementWiseLayer<_Functor>(_Functor()) ); }
@@ -2611,5 +2701,32 @@ Ptr<Layer> ChannelsPReLULayer::create(const LayerParams& params)
     return l;
 }
 
+Ptr<SignLayer> SignLayer::create(const LayerParams& params)
+{
+    Ptr<SignLayer> l(new ElementWiseLayer<SignFunctor>());
+    l->setParamsFrom(params);
+
+    return l;
+}
+
+Ptr<ReciprocalLayer> ReciprocalLayer::create(const LayerParams& params)
+{
+    Ptr<ReciprocalLayer> l(new ElementWiseLayer<ReciprocalFunctor>());
+    l->setParamsFrom(params);
+
+    return l;
+}
+
+Ptr<ShrinkLayer> ShrinkLayer::create(const LayerParams& params)
+{
+    float bias = params.get<float>("bias", 0.f);
+    float lambd = params.get<float>("lambd", 0.5f);
+    Ptr<ShrinkLayer> l(new ElementWiseLayer<ShrinkFunctor>(ShrinkFunctor(bias, lambd)));
+    l->setParamsFrom(params);
+    l->bias = bias;
+    l->lambd = lambd;
+
+    return l;
+}
 }
 }
diff --git a/modules/dnn/src/onnx/onnx_importer.cpp b/modules/dnn/src/onnx/onnx_importer.cpp
@@ -3675,8 +3675,8 @@ void ONNXImporter::buildDispatchMap_ONNX_AI(int opset_version)
 
     std::vector<std::string> simpleLayers{"Acos", "Acosh", "Asin", "Asinh", "Atan", "Atanh", "Ceil", "Celu", "Cos",
                                           "Cosh", "Dropout", "Erf", "Exp", "Floor", "HardSigmoid", "HardSwish",
-                                          "Identity", "Log", "Round", "Selu", "Sigmoid", "Sin", "Sinh", "Softmax",
-                                          "Softplus", "Softsign", "Sqrt", "Tan", "ThresholdedRelu"};
+                                          "Identity", "Log", "Round", "Reciprocal", "Selu", "Sign", "Sigmoid", "Sin", "Sinh", "Softmax",
+                                          "Softplus", "Softsign", "Shrink", "Sqrt", "Tan", "ThresholdedRelu"};
     for (const auto& name : simpleLayers)
     {
         dispatch[name] = &ONNXImporter::parseSimpleLayers;

diff --git a/modules/dnn/src/opencl/activations.cl b/modules/dnn/src/opencl/activations.cl
@@ -306,3 +306,26 @@ __kernel void ThresholdedReluForward(const int n, __global T* in, __global T* ou
     if(index < n)
         out[index] = (in[index] > alpha ? in[index] : 0.f);
 }
+
+__kernel void ShrinkForward(const int n, __global T* in, __global T* out,
+                            const KERNEL_ARG_DTYPE bias,
+                            const KERNEL_ARG_DTYPE lambd)
+{
+    int index = get_global_id(0);
+    if(index < n)
+        out[index] = in[index] < -lambd ? in[index] + bias : (in[index] > lambd ? in[index] - bias : 0.f);
+}
+
+__kernel void SignForward(const int n, __global T* in, __global T* out)
+{
+    int index = get_global_id(0);
+    if(index < n)
+        out[index] = in[index] > 0.f ? 1.0f : (in[index] < 0.f) ? -1.0f : 0.0f);
+}
+
+__kernel void ReciprocalForward(const int n, __global T* in, __global T* out)
+{
+    int index = get_global_id(0);
+    if(index < n)
+        out[index] = 1.0f/in[index];
+}
diff --git a/modules/dnn/test/test_onnx_conformance_layer_parser_denylist.inl.hpp b/modules/dnn/test/test_onnx_conformance_layer_parser_denylist.inl.hpp
@@ -337,8 +337,6 @@
 "test_range_float_type_positive_delta_expanded",
 "test_range_int32_type_negative_delta",
 "test_range_int32_type_negative_delta_expanded",
-"test_reciprocal",
-"test_reciprocal_example",
 "test_reduce_sum_default_axes_keepdims_example",
 "test_reduce_sum_default_axes_keepdims_random",
 "test_reduce_sum_do_not_keepdims_example",
@@ -479,9 +477,6 @@
 "test_shape_start_1_end_2",
 "test_shape_start_1_end_negative_1",
 "test_shape_start_negative_1",
-"test_shrink_hard",
-"test_shrink_soft",
-"test_sign",
 "test_simple_rnn_batchwise",
 "test_simple_rnn_defaults",
 "test_simple_rnn_with_initial_bias",