introduce SoftmaxV2 for axis param

src/CMakeLists.txt

@@ -135,6 +135,7 @@ ncnn_add_layer(Normalize)
 ncnn_add_layer(Permute)
 ncnn_add_layer(PriorBox)
 ncnn_add_layer(ConcatV2)
+ncnn_add_layer(SoftmaxV2)
 
 add_library(ncnn STATIC ${ncnn_SRCS})
 

src/layer/softmaxv2.cpp

@@ -0,0 +1,329 @@
+// Tencent is pleased to support the open source community by making ncnn available.
+//
+// Copyright (C) 2017 THL A29 Limited, a Tencent company. All rights reserved.
+//
+// Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
+// in compliance with the License. You may obtain a copy of the License at
+//
+// https://opensource.org/licenses/BSD-3-Clause
+//
+// Unless required by applicable law or agreed to in writing, software distributed
+// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
+// CONDITIONS OF ANY KIND, either express or implied. See the License for the
+// specific language governing permissions and limitations under the License.
+
+#include "softmaxv2.h"
+#include <float.h>
+#include <math.h>
+#include <algorithm>
+
+namespace ncnn {
+
+DEFINE_LAYER_CREATOR(SoftmaxV2)
+
+SoftmaxV2::SoftmaxV2()
+{
+    one_blob_only = true;
+    support_inplace = false;
+}
+
+#if NCNN_STDIO
+#if NCNN_STRING
+int SoftmaxV2::load_param(FILE* paramfp)
+{
+    int nscan = fscanf(paramfp, "%d", &axis);
+    if (nscan != 1)
+    {
+        fprintf(stderr, "SoftmaxV2 load_param failed %d\n", nscan);
+        return -1;
+    }
+
+    return 0;
+}
+#endif // NCNN_STRING
+int SoftmaxV2::load_param_bin(FILE* paramfp)
+{
+    fread(&axis, sizeof(int), 1, paramfp);
+
+    return 0;
+}
+#endif // NCNN_STDIO
+
+int SoftmaxV2::load_param(const unsigned char*& mem)
+{
+    axis = *(int*)(mem);
+    mem += 4;
+
+    return 0;
+}
+
+int SoftmaxV2::forward(const Mat& bottom_blob, Mat& top_blob) const
+{
+    // value = exp( value - global max value )
+    // sum all value
+    // value = value / sum
+
+    if (axis == 0)
+    {
+        int w = bottom_blob.w;
+        int h = bottom_blob.h;
+        int channels = bottom_blob.c;
+        int size = w * h;
+
+        top_blob.create(w, h, channels);
+        if (top_blob.empty())
+            return -100;
+
+        Mat max;
+        max.create(w, h);
+        if (max.empty())
+            return -100;
+        max.fill(-FLT_MAX);
+        for (int q=0; q<channels; q++)
+        {
+            const float* ptr = bottom_blob.channel(q);
+            float* maxptr = max;
+
+            for (int i=0; i<size; i++)
+            {
+                maxptr[i] = std::max(maxptr[i], ptr[i]);
+            }
+        }
+
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            const float* ptr = bottom_blob.channel(q);
+            float* outptr = top_blob.channel(q);
+            float* maxptr = max;
+
+            for (int i=0; i<size; i++)
+            {
+                outptr[i] = exp(ptr[i] - maxptr[i]);
+            }
+        }
+
+        Mat sum;
+        sum.create(w, h);
+        if (sum.empty())
+            return -100;
+        sum.fill(0.f);
+        for (int q=0; q<channels; q++)
+        {
+            const float* outptr = top_blob.channel(q);
+            float* sumptr = sum;
+
+            for (int i=0; i<size; i++)
+            {
+                sumptr[i] += outptr[i];
+            }
+        }
+
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            float* outptr = top_blob.channel(q);
+            float* sumptr = sum;
+
+            for (int i=0; i<size; i++)
+            {
+                outptr[i] /= sumptr[i];
+            }
+        }
+
+    }
+    else if (axis == 1)
+    {
+        int w = bottom_blob.w;
+        int h = bottom_blob.h;
+        int channels = bottom_blob.c;
+        int size = w * h;
+
+        top_blob.create(w, h, channels);
+        if (top_blob.empty())
+            return -100;
+
+        Mat max;
+        max.create(h, channels);
+        if (max.empty())
+            return -100;
+        max.fill(-FLT_MAX);
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            const float* ptr = bottom_blob.channel(q);
+            float* maxptr = max.row(q);
+
+            for (int i=0; i<h; i++)
+            {
+                float max = -FLT_MAX;
+                for (int j=0; j<w; j++)
+                {
+                    max = std::max(max, ptr[j]);
+                }
+
+                maxptr[i] = max;
+                ptr += w;
+            }
+        }
+
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            const float* ptr = bottom_blob.channel(q);
+            float* outptr = top_blob.channel(q);
+            float* maxptr = max.row(q);
+
+            for (int i=0; i<h; i++)
+            {
+                float max = maxptr[i];
+                for (int j=0; j<w; j++)
+                {
+                    outptr[j] = exp(ptr[j] - max);
+                }
+
+                ptr += w;
+                outptr += w;
+            }
+        }
+
+        Mat sum;
+        sum.create(h, channels);
+        if (sum.empty())
+            return -100;
+        sum.fill(0.f);
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            const float* outptr = top_blob.channel(q);
+            float* sumptr = sum.row(q);
+
+            for (int i=0; i<h; i++)
+            {
+                float sum = 0.f;
+                for (int j=0; j<w; j++)
+                {
+                    sum += outptr[j];
+                }
+
+                sumptr[i] = sum;
+                outptr += w;
+            }
+        }
+
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            float* outptr = top_blob.channel(q);
+            float* sumptr = sum.row(q);
+
+            for (int i=0; i<h; i++)
+            {
+                float sum = sumptr[i];
+                for (int j=0; j<w; j++)
+                {
+                    outptr[j] /= sum;
+                }
+
+                outptr += w;
+            }
+        }
+
+    }
+    else if (axis == 2)
+    {
+        int w = bottom_blob.w;
+        int h = bottom_blob.h;
+        int channels = bottom_blob.c;
+        int size = w * h;
+
+        top_blob.create(w, h, channels);
+        if (top_blob.empty())
+            return -100;
+
+        Mat max;
+        max.create(w, channels);
+        if (max.empty())
+            return -100;
+        max.fill(-FLT_MAX);
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            const float* ptr = bottom_blob.channel(q);
+            float* maxptr = max.row(q);
+
+            for (int i=0; i<h; i++)
+            {
+                for (int j=0; j<w; j++)
+                {
+                    maxptr[j] = std::max(maxptr[j], ptr[j]);
+                }
+
+                ptr += w;
+            }
+        }
+
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            const float* ptr = bottom_blob.channel(q);
+            float* outptr = top_blob.channel(q);
+            float* maxptr = max.row(q);
+
+            for (int i=0; i<h; i++)
+            {
+                for (int j=0; j<w; j++)
+                {
+                    outptr[j] = exp(ptr[j] - maxptr[j]);
+                }
+
+                ptr += w;
+                outptr += w;
+            }
+        }
+
+        Mat sum;
+        sum.create(w, channels);
+        if (sum.empty())
+            return -100;
+        sum.fill(0.f);
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            const float* outptr = top_blob.channel(q);
+            float* sumptr = sum.row(q);
+
+            for (int i=0; i<h; i++)
+            {
+                for (int j=0; j<w; j++)
+                {
+                    sumptr[j] += outptr[j];
+                }
+
+                outptr += w;
+            }
+        }
+
+        #pragma omp parallel for
+        for (int q=0; q<channels; q++)
+        {
+            float* outptr = top_blob.channel(q);
+            float* sumptr = sum.row(q);
+
+            for (int i=0; i<h; i++)
+            {
+                for (int j=0; j<w; j++)
+                {
+                    outptr[j] /= sumptr[j];
+                }
+
+                outptr += w;
+            }
+        }
+
+    }
+
+    return 0;
+}
+
+} // namespace ncnn

src/layer/softmaxv2.h

@@ -0,0 +1,43 @@
+// Tencent is pleased to support the open source community by making ncnn available.
+//
+// Copyright (C) 2017 THL A29 Limited, a Tencent company. All rights reserved.
+//
+// Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
+// in compliance with the License. You may obtain a copy of the License at
+//
+// https://opensource.org/licenses/BSD-3-Clause
+//
+// Unless required by applicable law or agreed to in writing, software distributed
+// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
+// CONDITIONS OF ANY KIND, either express or implied. See the License for the
+// specific language governing permissions and limitations under the License.
+
+#ifndef LAYER_SOFTMAXV2_H
+#define LAYER_SOFTMAXV2_H
+
+#include "layer.h"
+
+namespace ncnn {
+
+class SoftmaxV2 : public Layer
+{
+public:
+    SoftmaxV2();
+
+#if NCNN_STDIO
+#if NCNN_STRING
+    virtual int load_param(FILE* paramfp);
+#endif // NCNN_STRING
+    virtual int load_param_bin(FILE* paramfp);
+#endif // NCNN_STDIO
+    virtual int load_param(const unsigned char*& mem);
+
+    virtual int forward(const Mat& bottom_blob, Mat& top_blob) const;
+
+public:
+    int axis;
+};
+
+} // namespace ncnn
+
+#endif // LAYER_SOFTMAXV2_H