release preview4.

Author: Oceania2018

src/SciSharp.TensorFlow.Redist/README.md

@@ -27,6 +27,6 @@ Related merged [commits](https://github.com/SciSharp/TensorFlow.NET/commit/854a5
 On Windows, the tar command does not support extracting archives with symlinks. So when `dotnet pack` runs on Windows it will only package the Windows binaries.
 
 1. Run `dotnet pack SciSharp.TensorFlow.Redist.nupkgproj` under `src/SciSharp.TensorFlow.Redist` directory in Linux.
-2. Run `dotnet nuget push SciSharp.TensorFlow.Redist.1.15.0.nupkg -k APIKEY -s https://api.nuget.org/v3/index.json`
+2. Run `dotnet nuget push SciSharp.TensorFlow.Redist.2.3.0.nupkg -k APIKEY -s https://api.nuget.org/v3/index.json`
 
 

src/TensorFlowNET.Core/Data/DatasetV2.cs

@@ -95,7 +95,7 @@ public override string ToString()
 
         public IEnumerator<(Tensor, Tensor)> GetEnumerator()
         {
-            var ownedIterator = new OwnedIterator(this);
+            using var ownedIterator = new OwnedIterator(this);
 
             Tensor[] results = null;
             while (true)

src/TensorFlowNET.Core/Data/OwnedIterator.cs

@@ -8,7 +8,7 @@ namespace Tensorflow
     /// <summary>
     /// An iterator producing tf.Tensor objects from a tf.data.Dataset.
     /// </summary>
-    public class OwnedIterator : IteratorBase
+    public class OwnedIterator : IteratorBase, IDisposable
     {
         IDatasetV2 _dataset;
         TensorSpec[] _element_spec;
@@ -45,5 +45,10 @@ public Tensor[] next()
                 throw new StopIteration(ex.Message);
             }
         }
+
+        public void Dispose()
+        {
+            _resource_deleter.Dispose();
+        }
     }
 }

src/TensorFlowNET.Core/Keras/KerasApi.cs

@@ -17,7 +17,7 @@ public class KerasApi
         public Initializers initializers { get; } = new Initializers();
         public LayersApi layers { get; } = new LayersApi();
         public Activations activations { get; } = new Activations();
-
+        public Preprocessing preprocessing { get; } = new Preprocessing();
         public BackendImpl backend { get; } = new BackendImpl();
 
         public Sequential Sequential(List<Layer> layers = null,

src/TensorFlowNET.Core/Keras/Preprocessing.cs

@@ -0,0 +1,11 @@
+using System;
+using System.Collections.Generic;
+using System.Text;
+
+namespace Tensorflow.Keras.Preprocessings
+{
+    public partial class DatasetUtils
+    {
+
+    }
+}

src/TensorFlowNET.Core/Keras/Preprocessings/DatasetUtils.cs

@@ -0,0 +1,33 @@
+using System;
+using System.Collections.Generic;
+using System.Text;
+
+namespace Tensorflow.Keras.Preprocessings
+{
+    public partial class DatasetUtils
+    {
+        /// <summary>
+        /// Make list of all files in the subdirs of `directory`, with their labels.
+        /// </summary>
+        /// <param name="directory"></param>
+        /// <param name="labels"></param>
+        /// <param name="formats"></param>
+        /// <param name="class_names"></param>
+        /// <param name="shuffle"></param>
+        /// <param name="seed"></param>
+        /// <param name="follow_links"></param>
+        /// <returns>
+        /// file_paths, labels, class_names
+        /// </returns>
+        public (string[], int[], string[]) index_directory(string directory,
+            string labels,
+            string[] formats,
+            string class_names = null,
+            bool shuffle = true,
+            int? seed = null,
+            bool follow_links = false)
+        {
+            throw new NotImplementedException("");
+        }
+    }
+}

src/TensorFlowNET.Core/Keras/Preprocessings/DatasetUtils.index_directory.cs

@@ -0,0 +1,10 @@
+using Tensorflow.Keras.Preprocessings;
+
+namespace Tensorflow.Keras
+{
+    public partial class Preprocessing
+    {
+        public Sequence sequence => new Sequence();
+        public DatasetUtils dataset_utils => new DatasetUtils();
+    }
+}

src/TensorFlowNET.Core/Keras/Preprocessings/Preprocessing.cs

@@ -0,0 +1,57 @@
+using System;
+using static Tensorflow.Binding;
+
+namespace Tensorflow.Keras
+{
+    public partial class Preprocessing
+    {
+        public static string[] WHITELIST_FORMATS = new[] { ".bmp", ".gif", ".jpeg", ".jpg", ".png" };
+
+        /// <summary>
+        /// Generates a `tf.data.Dataset` from image files in a directory.
+        /// https://www.tensorflow.org/api_docs/python/tf/keras/preprocessing/image_dataset_from_directory
+        /// </summary>
+        /// <param name="directory">Directory where the data is located.</param>
+        /// <param name="labels"></param>
+        /// <param name="label_mode"></param>
+        /// <param name="class_names"></param>
+        /// <param name="color_mode"></param>
+        /// <param name="batch_size"></param>
+        /// <param name="image_size"></param>
+        /// <param name="shuffle"></param>
+        /// <param name="seed"></param>
+        /// <param name="validation_split"></param>
+        /// <param name="subset"></param>
+        /// <param name="interpolation"></param>
+        /// <param name="follow_links"></param>
+        /// <returns></returns>
+        public Tensor image_dataset_from_directory(string directory,
+            string labels = "inferred",
+            string label_mode = "int",
+            string class_names = null,
+            string color_mode = "rgb",
+            int batch_size = 32,
+            TensorShape image_size = null,
+            bool shuffle = true,
+            int? seed = null,
+            float validation_split = 0.2f,
+            string subset = null,
+            string interpolation = "bilinear",
+            bool follow_links = false)
+        {
+            int num_channels = 0;
+            if (color_mode == "rgb")
+                num_channels = 3;
+            // C:/Users/haipi/.keras/datasets/flower_photos
+            var (image_paths, label_list, class_name_list) = tf.keras.preprocessing.dataset_utils.index_directory(directory,
+                labels,
+                WHITELIST_FORMATS,
+                class_names: class_names,
+                shuffle: shuffle,
+                seed: seed,
+                follow_links: follow_links);
+
+            throw new NotImplementedException("");
+        }
+    }
+}

src/TensorFlowNET.Core/Keras/Preprocessings/Preprocessing.image_dataset_from_directory.cs

@@ -37,7 +37,7 @@ PM> Install-Package SciSharp.TensorFlow.Redist-Linux-GPU
 
 ### Download prebuild binary manually
 
-We can't found official prebuild binaries for each platform since tensorflow 2.0. If you know where we can download, please PR here.
+Tensorflow packages are built nightly and uploaded to GCS for all supported platforms. They are uploaded to the [libtensorflow-nightly](https://www.tensorflow.org/install/lang_c) GCS bucket and are indexed by operating system and date built.
 
 
 ### Build from source for Windows