Issue with ImageLoader working with empty input column

inputfile.tsv

test.cs

using Microsoft.ML;
using Microsoft.ML.Data;

namespace TestNugets
{
    class Test5
    {
        public static void Example()
        {
            var inputFilePath = @"C:\Users\anvelazq\Desktop\inputfile.tsv";
            var inputFolder = @"C:\I-dont-exist";

            MLContext mlContext = new MLContext(seed: 1);

            // Because of a typo on ModelInput class, this dataview won't load
            // the values in the ImagePath column, but the program would still run without loading images
            IDataView trainingDataView = mlContext.Data.LoadFromTextFile<ModelInput>(
                                path: inputFilePath,
                                hasHeader: true,
                                separatorChar: '\t',
                                allowQuoting: true,
                                allowSparse: false);

            // This DataView will have empty values in its ImagePath column
            // but the program would still run without loading images
            IDataView trainingDataView2 = mlContext.Data.LoadFromEnumerable<ModelInput>(
                    new ModelInput[]
                    {
                        new ModelInput { Label = "dog", ImagePath = null },
                        new ModelInput { Label = "cat" },
                        new ModelInput { Label = "elephant", ImagePath = ""},
                        // new ModelInput { Label = "snake", ImagePath = "snake.bmp"} // if uncommented, this program correctly throws an exception because the inputFolder doesn't exist
                    }
                );

            var dataProcessPipeline = mlContext.Transforms.Conversion.MapValueToKey("Label", "Label")
                          .Append(mlContext.Transforms.LoadImages("ImagePath_featurized", inputFolder, "ImagePath"))
                          .Append(mlContext.Transforms.ResizeImages("ImagePath_featurized", 224, 224, "ImagePath_featurized"))
                          .Append(mlContext.Transforms.ExtractPixels("ImagePath_featurized", "ImagePath_featurized"))
                          .Append(mlContext.Transforms.DnnFeaturizeImage("ImagePath_featurized", m => m.ModelSelector.ResNet18(mlContext, m.OutputColumn, m.InputColumn), "ImagePath_featurized"))
                          .Append(mlContext.Transforms.Concatenate("Features", new[] { "ImagePath_featurized" }))
                          .Append(mlContext.Transforms.NormalizeMinMax("Features", "Features"))
                          .AppendCacheCheckpoint(mlContext);

            var trainer = mlContext.MulticlassClassification.Trainers.OneVersusAll(mlContext.BinaryClassification.Trainers.AveragedPerceptron(labelColumnName: "Label", numberOfIterations: 10, featureColumnName: "Features"), labelColumnName: "Label")
                                      .Append(mlContext.Transforms.Conversion.MapKeyToValue("PredictedLabel", "PredictedLabel"));
            var trainingPipeline = dataProcessPipeline.Append(trainer);

            // Fit and transform with DataView 1
            ITransformer model = trainingPipeline.Fit(trainingDataView);
            var transformedDataView = model.Transform(trainingDataView);

            // Fit and transform with DataView 2
            ITransformer model2 = trainingPipeline.Fit(trainingDataView2);
            var transformedDataView2 = model2.Transform(trainingDataView2);
            
            // In both options no images were loaded, but still the pipeline assigned PredictedLabels to each row
        }

        class ModelInput
        {
            [ColumnName("Label"), LoadColumn(0)]
            public string Label { get; set; }

            [ColumnName("ImagePath"), LoadColumn(2)] // Column 2 doesnt exist in the input file, so it doesnt load anything into ImagePath
            public string ImagePath { get; set; }
        }
    }
}