Skip to content

Retro-fitted changes made to the ML docs repo into ML.NET #4457

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
codemzs merged 1 commit into from
Nov 8, 2019
Merged

Retro-fitted changes made to the ML docs repo into ML.NET #4457

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
39 changes: 19 additions & 20 deletions src/Microsoft.ML.Vision/ImageClassificationTrainer.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,10 +42,10 @@ namespace Microsoft.ML.Vision
/// </summary>
/// <remarks>
/// <format type="text/markdown"><![CDATA[
/// To create this trainer, use [ImageClassification](xref:Microsoft.ML.Vision.DnnCatalog.ImageClassification(Microsoft.ML.MulticlassClassificationCatalog.MulticlassClassificationTrainers,System.String,System.String,System.String,System.String,Microsoft.ML.IDataView)).
/// To create this trainer, use [ImageClassification](xref:Microsoft.ML.VisionCatalog.ImageClassification(Microsoft.ML.MulticlassClassificationCatalog.MulticlassClassificationTrainers,System.String,System.String,System.String,System.String,Microsoft.ML.IDataView)).
///
/// ### Input and Output Columns
/// The input label column data must be[key] (xref:Microsoft.ML.Data.KeyDataViewType) type and the feature column must be a variable-sized vector of<xref:System.Byte>.
/// The input label column data must be [key](xref:Microsoft.ML.Data.KeyDataViewType) type and the feature column must be a variable-sized vector of <xref:System.Byte>.
///
/// This trainer outputs the following columns:
///
Expand All @@ -66,8 +66,7 @@ namespace Microsoft.ML.Vision
///
/// ### Training Algorithm Details
/// Trains a Deep Neural Network(DNN) by leveraging an existing pre-trained model such as Resnet50 for the purpose
/// of classifying images. The technique was inspired from [TensorFlow's retrain image classification tutorial]
/// (https://www.tensorflow.org/hub/tutorials/image_retraining)
/// of classifying images. The technique was inspired from [TensorFlow's retrain image classification tutorial](https://www.tensorflow.org/hub/tutorials/image_retraining)
/// ]]>
/// </format>
/// </remarks>
Expand Down Expand Up @@ -339,79 +338,79 @@ public enum Dataset
public sealed class Options : TrainerInputBaseWithLabel
{
/// <summary>
/// Number of samples to use for mini-batch training.
/// Number of samples to use for mini-batch training. The default value for BatchSize is 10.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Number of samples to use for mini-batch training.", SortOrder = 9)]
public int BatchSize = 10;

/// <summary>
/// Number of training iterations.
/// Number of training iterations. The default value for Epoch is 200.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Number of training iterations.", SortOrder = 10)]
public int Epoch = 200;

/// <summary>
/// Learning rate to use during optimization.
/// Learning rate to use during optimization. The default value for Learning Rate is 0.01.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Learning rate to use during optimization.", SortOrder = 12)]
public float LearningRate = 0.01f;

/// <summary>
/// Early stopping technique parameters to be used to terminate training when training metric stops improving.
/// Early stopping technique parameters to be used to terminate training when training metric stops improving. By default EarlyStopping is turned on and the monitoring metric is Accuracy.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Early stopping technique parameters to be used to terminate training when training metric stops improving.", SortOrder = 15)]
public EarlyStopping EarlyStoppingCriteria = new EarlyStopping();

/// <summary>
/// Specifies the model architecture to be used in the case of image classification training using transfer learning.
/// Specifies the model architecture to be used in the case of image classification training using transfer learning. The default Architecture is Resnet_v2_50.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Model architecture to be used in transfer learning for image classification.", SortOrder = 15)]
public Architecture Arch = Architecture.ResnetV250;

/// <summary>
/// Name of the tensor that will contain the output scores of the last layer when transfer learning is done.
/// Name of the tensor that will contain the output scores of the last layer when transfer learning is done. The default tensor name is "Score".
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Softmax tensor of the last layer in transfer learning.", SortOrder = 15)]
public string ScoreColumnName = "Score";

/// <summary>
/// Name of the tensor that will contain the predicted label from output scores of the last layer when transfer learning is done.
/// Name of the tensor that will contain the predicted label from output scores of the last layer when transfer learning is done. The default tensor name is "PredictedLabel".
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Argmax tensor of the last layer in transfer learning.", SortOrder = 15)]
public string PredictedLabelColumnName = "PredictedLabel";

/// <summary>
/// Final model and checkpoint files/folder prefix for storing graph files.
/// Final model and checkpoint files/folder prefix for storing graph files. The default prefix is "custom_retrained_model_based_on_".
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Final model and checkpoint files/folder prefix for storing graph files.", SortOrder = 15)]
public string FinalModelPrefix = "custom_retrained_model_based_on_";

/// <summary>
/// Callback to report statistics on accuracy/cross entropy during training phase.
/// Callback to report statistics on accuracy/cross entropy during training phase. Metrics Callback is set to null by default.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Callback to report metrics during training and validation phase.", SortOrder = 15)]
public Action<ImageClassificationMetrics> MetricsCallback = null;

/// <summary>
/// Indicates the path where the image bottleneck cache files and trained model are saved, default is a new temporary directory
/// Indicates the path where the image bottleneck cache files and trained model are saved, default is a new temporary directory.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Indicates the path where the models get downloaded to and cache files saved, default is a new temporary directory.", SortOrder = 15)]
public string WorkspacePath = null;

/// <summary>
/// Indicates to evaluate the model on train set after every epoch.
/// Indicates to evaluate the model on train set after every epoch. Test on trainset is set to true by default.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Indicates to evaluate the model on train set after every epoch.", SortOrder = 15)]
public bool TestOnTrainSet = true;

/// <summary>
/// Indicates to not re-compute cached bottleneck trainset values if already available in the bin folder.
/// Indicates to not re-compute cached bottleneck trainset values if already available in the bin folder. This parameter is set to false by default.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Indicates to not re-compute trained cached bottleneck values if already available in the bin folder.", SortOrder = 15)]
public bool ReuseTrainSetBottleneckCachedValues = false;

/// <summary>
/// Indicates to not re-compute cached bottleneck validationset values if already available in the bin folder.
/// Indicates to not re-compute cached bottleneck validationset values if already available in the bin folder. This parameter is set to false by default.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Indicates to not re-compute validataionset cached bottleneck validationset values if already available in the bin folder.", SortOrder = 15)]
public bool ReuseValidationSetBottleneckCachedValues = false;
Expand All @@ -423,19 +422,19 @@ public sealed class Options : TrainerInputBaseWithLabel
public IDataView ValidationSet;

/// <summary>
/// Indicates the file name within the workspace to store trainset bottleneck values for caching.
/// Indicates the file name within the workspace to store trainset bottleneck values for caching, default file name is "trainSetBottleneckFile.csv".
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Indicates the file name to store trainset bottleneck values for caching.", SortOrder = 15)]
public string TrainSetBottleneckCachedValuesFileName = "trainSetBottleneckFile.csv";

/// <summary>
/// Indicates the file name within the workspace to store validationset bottleneck values for caching.
/// Indicates the file name within the workspace to store validationset bottleneck values for caching, default file name is "validationSetBottleneckFile.csv".
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "Indicates the file name to store validationset bottleneck values for caching.", SortOrder = 15)]
public string ValidationSetBottleneckCachedValuesFileName = "validationSetBottleneckFile.csv";

/// <summary>
/// A class that performs learning rate scheduling.
/// A class that performs learning rate scheduling. The default learning rate scheduler is exponential learning rate decay.
/// </summary>
[Argument(ArgumentType.AtMostOnce, HelpText = "A class that performs learning rate scheduling.", SortOrder = 15)]
public LearningRateScheduler LearningRateScheduler = new ExponentialLRDecay();
Expand Down
46 changes: 25 additions & 21 deletions src/Microsoft.ML.Vision/VisionCatalog.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,12 @@

namespace Microsoft.ML
{
/// <summary>
/// Collection of extension methods for <see cref="T:Microsoft.ML.MulticlassClassificationCatalog.MulticlassClassificationTrainers" /> to create instances of ImageClassification trainer components.
/// </summary>
/// <remarks>
/// This requires additional nuget dependencies to link against Tensorflow native dlls. See <see cref="T:Microsoft.ML.Vision.ImageClassificationTrainer"/> for more information.
/// </remarks>
public static class VisionCatalog
{

Expand Down Expand Up @@ -79,40 +85,38 @@ internal static DnnRetrainEstimator RetrainDnnModel(
}

/// <summary>
/// Performs image classification using transfer learning.
/// Usage of this API requires additional NuGet dependencies on TensorFlow redist, see linked document
/// for more information.
/// <format type="text/markdown">
/// <![CDATA[
/// [!include[io](~/../docs/samples/docs/api-reference/tensorflow-usage.md)]
/// ]]>
/// </format>
/// Create <see cref="Microsoft.ML.Vision.ImageClassificationTrainer"/> using advanced options, which trains a Deep Neural Network(DNN) to classify images.
/// </summary>
/// <param name="catalog">Catalog</param>
/// <param name="options">An <see cref="ImageClassificationTrainer.Options"/> object specifying advanced
/// options for <see cref="ImageClassificationTrainer"/>.</param>
/// <example>
/// <format type="text/markdown">
/// <![CDATA[
/// [!code-csharp[ImageClassification](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/MulticlassClassification/ImageClassification/ResnetV2101TransferLearningTrainTestSplit.cs)]
/// ]]></format>
/// </example>

public static ImageClassificationTrainer ImageClassification(
this MulticlassClassificationCatalog.MulticlassClassificationTrainers catalog,
ImageClassificationTrainer.Options options) =>
new ImageClassificationTrainer(CatalogUtils.GetEnvironment(catalog), options);

/// <summary>
/// Performs image classification using transfer learning.
/// Usage of this API requires additional NuGet dependencies on TensorFlow redist, see linked document for
/// more information.
/// <format type="text/markdown">
/// <![CDATA[
/// [!include[io](~/../docs/samples/docs/api-reference/tensorflow-usage.md)]
/// ]]>
/// </format>
/// Create <see cref="Microsoft.ML.Vision.ImageClassificationTrainer"/>, which trains a Deep Neural Network(DNN) to classify images.
/// </summary>
/// <param name="catalog">Catalog</param>
/// <param name="labelColumnName">The name of the labels column.</param>
/// <param name="featureColumnName">The name of the input features column.</param>
/// <param name="scoreColumnName">The name of the output score column.</param>
/// <param name="predictedLabelColumnName">The name of the output predicted label columns.</param>
/// <param name="validationSet">The validation set used while training to improve model quality.</param>
/// <param name="labelColumnName">The name of the labels column. The default for this parameter is "label".</param>
/// <param name="featureColumnName">The name of the input features column. The default for this parameter is "Features".</param>
/// <param name="scoreColumnName">The name of the output score column. The default for this parameter is "Score"</param>
/// <param name="predictedLabelColumnName">The name of the output predicted label columns. The default for this parameter is "PredictedLabel"</param>
/// <param name="validationSet">The validation set used while training to improve model quality. The default for this parameter is null.</param>
/// <example>
/// <format type="text/markdown">
/// <![CDATA[
/// [!code-csharp[ImageClassification](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/MulticlassClassification/ImageClassification/ImageClassificationDefault.cs)]
/// ]]></format>
/// </example>

public static ImageClassificationTrainer ImageClassification(
this MulticlassClassificationCatalog.MulticlassClassificationTrainers catalog,
Expand Down