Modified how DataViewTypes are registered

test/Microsoft.ML.Tests/Microsoft.ML.Tests.csproj

@@ -54,4 +54,9 @@
     <PackageReference Include="Microsoft.ML.TestModels" Version="$(MicrosoftMLTestModelsPackageVersion)" />
     <PackageReference Include="System.Data.SqlClient" Version="$(SystemDataSqlClientVersion)" />
   </ItemGroup>
+  <ItemGroup>
+    <None Update="column_name_test\model.onnx">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </None>
+  </ItemGroup>
 </Project>

test/Microsoft.ML.Tests/OnnxSequenceTypeWithAttributesTest.cs

@@ -0,0 +1,71 @@
+using System;
+using System.Collections.Generic;
+using System.Drawing;
+using System.Text;
+using Microsoft.ML.Data;
+using Microsoft.ML.RunTests;
+using Microsoft.ML.Transforms.Image;
+using Microsoft.ML.Transforms.Onnx;
+using Xunit;
+using Xunit.Abstractions;
+using System.Linq;
+using System.IO;
+
+namespace Microsoft.ML.Tests
+{
+    public class OnnxSequenceTypeWithAttributesTest : BaseTestBaseline
+    {
+        public class ImagePrediction
+        {
+            [ColumnName("classLabel")]
+            [VectorType]
+            public string[] Prediction;
+
+            [ColumnName("loss")]
+            [OnnxSequenceType(typeof(IDictionary<string, float>))]
+            public IEnumerable<IDictionary<string, float>> Loss;
+        }
+        public class ImageInput
+        {
+            [ImageType(224, 224)]
+            public Bitmap Image { get; set; }
+        }
+
+        public OnnxSequenceTypeWithAttributesTest(ITestOutputHelper output) : base(output)
+        {
+        }
+        public static PredictionEngine<ImageInput, ImagePrediction> LoadModel(string onnxModelFilePath)
+        {
+            var ctx = new MLContext();
+            var dataView = ctx.Data.LoadFromEnumerable(new List<ImageInput>());
+
+            var pipeline = ctx.Transforms.ResizeImages(
+                                resizing: ImageResizingEstimator.ResizingKind.Fill,
+                                outputColumnName: "data",
+                                imageWidth: 224,
+                                imageHeight: 224,
+                                inputColumnName: nameof(ImageInput.Image))
+                            .Append(ctx.Transforms.ExtractPixels(outputColumnName: "data"))
+                            .Append(ctx.Transforms.ApplyOnnxModel(
+                                modelFile: onnxModelFilePath,
+                                outputColumnNames: new[] { "classLabel", "loss" }, inputColumnNames: new[] { "data" }));
+
+            var model = pipeline.Fit(dataView);
+            return ctx.Model.CreatePredictionEngine<ImageInput, ImagePrediction>(model);
+        }
+
+        [Fact]
+        public void OnnxSequenceTypeWithColumnNameAttributeTest()
+        {
+            var modelFile = @"column_name_test/model.onnx";
+            var predictor = LoadModel(modelFile);
+            string image_path = Path.Combine(DataDir, "images", "banana.jpg");
+
+            var output = predictor.Predict(new ImageInput { Image = (Bitmap)Image.FromFile(image_path) });
+            Assert.NotEmpty(output.Prediction);
+            var loss = output.Loss.FirstOrDefault();
+            Assert.NotEmpty(loss);
+            Assert.True(loss[output.Prediction[0]] > 0, "Invalid output");
+        }
+    }
+}