Fix binary_accuracy for keras.

src/TensorFlowNET.Core/Tensorflow.Binding.csproj

@@ -5,7 +5,7 @@
     <AssemblyName>TensorFlow.NET</AssemblyName>
     <RootNamespace>Tensorflow</RootNamespace>
     <TargetTensorFlow>2.2.0</TargetTensorFlow>
-    <Version>0.40.0</Version>
+    <Version>0.40.1</Version>
     <LangVersion>8.0</LangVersion>
     <Authors>Haiping Chen, Meinrad Recheis, Eli Belash</Authors>
     <Company>SciSharp STACK</Company>
@@ -19,7 +19,7 @@
     <Description>Google's TensorFlow full binding in .NET Standard.
 Building, training and infering deep learning models.
 https://tensorflownet.readthedocs.io</Description>
-    <AssemblyVersion>0.40.0.0</AssemblyVersion>
+    <AssemblyVersion>0.40.1.0</AssemblyVersion>
     <PackageReleaseNotes>tf.net 0.20.x and above are based on tensorflow native 2.x.
 
 * Eager Mode is added finally.
@@ -32,7 +32,7 @@ TensorFlow .NET v0.3x is focused on making more Keras API works.
 Keras API is a separate package released as TensorFlow.Keras.
 
 tf.net 0.4x.x aligns with TensorFlow v2.4.1 native library.</PackageReleaseNotes>
-    <FileVersion>0.40.0.0</FileVersion>
+    <FileVersion>0.40.1.0</FileVersion>
     <PackageLicenseFile>LICENSE</PackageLicenseFile>
     <PackageRequireLicenseAcceptance>true</PackageRequireLicenseAcceptance>
     <SignAssembly>true</SignAssembly>

src/TensorFlowNET.Core/Tensors/tensor_util.cs

@@ -596,8 +596,6 @@ would not be rank 1.", tensor.op.get_attr("axis")));
                 case TF_DataType.TF_STRING:
                     return string.Join(string.Empty, nd.ToArray<byte>()
                         .Select(x => x < 32 || x > 127 ? "\\x" + x.ToString("x") : Convert.ToChar(x).ToString()));
-                case TF_DataType.TF_BOOL:
-                    return nd.GetBoolean(0).ToString();
                 case TF_DataType.TF_VARIANT:
                 case TF_DataType.TF_RESOURCE:
                     return "<unprintable>";

src/TensorFlowNET.Keras/BackendImpl.cs

@@ -137,6 +137,14 @@ namespace Tensorflow.Keras
         {
             _MANUAL_VAR_INIT = value;
         }
+
+        public Tensor mean(Tensor x, int axis = -1, bool keepdims = false)
+        {
+            if (x.dtype.as_base_dtype() == TF_DataType.TF_BOOL)
+                x = math_ops.cast(x, TF_DataType.TF_FLOAT);
+            return math_ops.reduce_mean(x, axis: new[] { axis }, keepdims: false);
+        }
+
         public GraphLearningPhase learning_phase()
         {
             var graph = tf.get_default_graph();

src/TensorFlowNET.Keras/Engine/MetricsContainer.cs

@@ -68,7 +68,9 @@ namespace Tensorflow.Keras.Engine
                 bool is_binary = y_p_last_dim == 1;
                 bool is_sparse_categorical = (y_t_rank < y_p_rank || y_t_last_dim == 1) && y_p_last_dim > 1;
 
-                if (is_sparse_categorical)
+                if (is_binary)
+                    metric_obj = keras.metrics.binary_accuracy;
+                else if (is_sparse_categorical)
                     metric_obj = keras.metrics.sparse_categorical_accuracy;
                 else
                     metric_obj = keras.metrics.categorical_accuracy;

src/TensorFlowNET.Keras/Metrics/MetricsApi.cs

@@ -1,7 +1,16 @@
-namespace Tensorflow.Keras.Metrics
+using static Tensorflow.KerasApi;
+
+namespace Tensorflow.Keras.Metrics
 {
     public class MetricsApi
     {
+        public Tensor binary_accuracy(Tensor y_true, Tensor y_pred)
+        {
+            float threshold = 0.5f;
+            y_pred = math_ops.cast(y_pred > threshold, y_pred.dtype);
+            return keras.backend.mean(math_ops.equal(y_true, y_pred), axis: -1);
+        }
+
         public Tensor categorical_accuracy(Tensor y_true, Tensor y_pred)
         {
             var eql = math_ops.equal(math_ops.argmax(y_true, -1), math_ops.argmax(y_pred, -1));

src/TensorFlowNET.Keras/Tensorflow.Keras.csproj

@@ -6,10 +6,10 @@
     <LangVersion>8.0</LangVersion>
     <RootNamespace>Tensorflow.Keras</RootNamespace>
     <Platforms>AnyCPU;x64</Platforms>
-    <Version>0.5.0</Version>
+    <Version>0.5.1</Version>
     <Authors>Haiping Chen</Authors>
     <Product>Keras for .NET</Product>
-    <Copyright>Apache 2.0, Haiping Chen 2020</Copyright>
+    <Copyright>Apache 2.0, Haiping Chen 2021</Copyright>
     <PackageId>TensorFlow.Keras</PackageId>
     <PackageProjectUrl>https://github.com/SciSharp/TensorFlow.NET</PackageProjectUrl>
     <PackageIconUrl>https://avatars3.githubusercontent.com/u/44989469?s=200&amp;v=4</PackageIconUrl>
@@ -35,8 +35,8 @@ Keras is an API designed for human beings, not machines. Keras follows best prac
     <RepositoryType>Git</RepositoryType>
     <SignAssembly>true</SignAssembly>
     <AssemblyOriginatorKeyFile>Open.snk</AssemblyOriginatorKeyFile>
-    <AssemblyVersion>0.5.0.0</AssemblyVersion>
-    <FileVersion>0.5.0.0</FileVersion>
+    <AssemblyVersion>0.5.1.0</AssemblyVersion>
+    <FileVersion>0.5.1.0</FileVersion>
     <PackageLicenseFile>LICENSE</PackageLicenseFile>
   </PropertyGroup>