[SAMPLES] YOLO Object Detection + Combo sample to enable Multiperson HRNet pose

AIDevGallery/Samples/ModelsDefinitions/imagemodels.modelgroup.json

@@ -16,8 +16,11 @@
             "Name": "Faster RCNN 10",
             "Url": "https://github.com/onnx/models/blob/main/validated/vision/object_detection_segmentation/faster-rcnn/model/FasterRCNN-10.onnx",
             "Description": "FPN",
-            "HardwareAccelerator": [ "CPU", "DML" ],
-            "SupportedOnQualcomm":  true,
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
+            "SupportedOnQualcomm": true,
             "Size": 167330019,
             "License": "mit"
           },
@@ -26,7 +29,10 @@
             "Name": "Faster RCNN 12",
             "Url": "https://github.com/onnx/models/blob/main/validated/vision/object_detection_segmentation/faster-rcnn/model/FasterRCNN-12.onnx",
             "Description": "FPN-fp32",
-            "HardwareAccelerator": [ "CPU", "DML" ],
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
             "SupportedOnQualcomm": true,
             "Size": 176713194,
             "License": "mit"
@@ -45,7 +51,10 @@
             "Name": "ResNet101 v1 7",
             "Url": "https://github.com/onnx/models/blob/main/validated/vision/classification/resnet/model/resnet101-v1-7.onnx",
             "Description": "ResNet101 v1 7",
-            "HardwareAccelerator": [ "CPU", "DML" ],
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
             "SupportedOnQualcomm": true,
             "Size": 178914043,
             "License": "apache-2.0"
@@ -55,7 +64,10 @@
             "Name": "ResNet50 v1 7",
             "Url": "https://github.com/onnx/models/blob/main/validated/vision/classification/resnet/model/resnet50-v1-7.onnx",
             "Description": "ResNet50 v1 7",
-            "HardwareAccelerator": [ "CPU", "DML" ],
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
             "SupportedOnQualcomm": true,
             "Size": 102583340,
             "License": "apache-2.0"
@@ -74,8 +86,11 @@
             "Name": "MobileNet v2 1.0",
             "Url": "https://github.com/onnx/models/blob/main/validated/vision/classification/mobilenet/model/mobilenetv2-10.onnx",
             "Description": "MobileNet v2 1.0",
-            "HardwareAccelerator": [ "CPU", "DML" ],
-            "SupportedOnQualcomm":  true,
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
+            "SupportedOnQualcomm": true,
             "Size": 13963115,
             "License": "apache-2.0"
           }
@@ -93,7 +108,10 @@
             "Name": "SqueezeNet 1.1",
             "Url": "https://github.com/onnx/models/blob/main/validated/vision/classification/squeezenet/model/squeezenet1.1-7.onnx",
             "Description": "SqueezeNet 1.1",
-            "HardwareAccelerator": [ "CPU", "DML" ],
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
             "SupportedOnQualcomm": true,
             "Size": 4956208,
             "License": "apache-2.0"
@@ -134,7 +152,10 @@
             "Name": "ESRGAN",
             "Url": "https://huggingface.co/microsoft/dml-ai-hub-models/blob/main/esrgan/esrgan.onnx",
             "Description": "ESRGAN",
-            "HardwareAccelerator": [ "CPU", "DML" ],
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
             "SupportedOnQualcomm": true,
             "Size": 67162397,
             "License": "apache-2.0"
@@ -153,7 +174,10 @@
             "Name": "FFNet 78s",
             "Url": "https://huggingface.co/microsoft/dml-ai-hub-models/blob/main/ffnet_78s/ffnet_78s.onnx",
             "Description": "FFNet 78s",
-            "HardwareAccelerator": [ "CPU", "DML" ],
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
             "SupportedOnQualcomm": true,
             "Size": 109974166,
             "License": "bsd-3-clause"
@@ -163,13 +187,38 @@
             "Name": "FFNet 54s",
             "Url": "https://huggingface.co/microsoft/dml-ai-hub-models/blob/main/ffnet_54s/ffnet_54s.onnx",
             "Description": "FFNet 54s",
-            "HardwareAccelerator": [ "CPU", "DML" ],
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
             "SupportedOnQualcomm": true,
             "Size": 72188137,
             "License": "bsd-3-clause"
           }
         },
         "ReadmeUrl": "https://huggingface.co/qualcomm/FFNet-78S/blob/main/README.md"
+      },
+      "YOLO": {
+        "Id": "yolo",
+        "Name": "YOLO",
+        "Description": "YOLO (You Only Look Once) is a state-of-the-art, real-time object detection algorithm.",
+        "DocsUrl": "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/yolov4",
+        "Models": {
+          "YOLOv4": {
+            "Id": "b7cae3a2-478f-1177-9559-709df5dfe703",
+            "Name": "YOLOv4",
+            "Url": "https://github.com/onnx/models/blob/main/validated/vision/object_detection_segmentation/yolov4/model/yolov4.onnx",
+            "Description": "YOLOv4 optimizes the speed and accuracy of object detection.",
+            "HardwareAccelerator": [
+              "CPU",
+              "DML"
+            ],
+            "SupportedOnQualcomm": true,
+            "Size": 257470589,
+            "License": "mit"
+          }
+        },
+        "ReadmeUrl": "https://github.com/onnx/models/blob/main/validated/vision/object_detection_segmentation/yolov4/README.md"
       }
     }
   }

AIDevGallery/Samples/Open Source Models/Image Models/ESRGAN/SuperResolution.xaml.cs

@@ -154,7 +154,7 @@ private async Task EnhanceImage(string filePath)
                 return finalOutputBitmap;
             });
 
-            BitmapImage outputImage = BitmapFunctions.ConvertBitmapToBitmapImageAsync(bitmapOutput);
+            BitmapImage outputImage = BitmapFunctions.ConvertBitmapToBitmapImage(bitmapOutput);
             NarratorHelper.AnnounceImageChanged(DefaultImage, "Image enhancement complete.");  // <exclude-line>
 
             bitmapOutput.Dispose();

AIDevGallery/Samples/Open Source Models/Image Models/FFNet/SegmentStreets.xaml.cs

@@ -237,7 +237,7 @@ private async Task Segment(string filePath)
             });
 
             // Convert the final overlay to BitmapImage for display
-            BitmapImage outputImage = BitmapFunctions.ConvertBitmapToBitmapImageAsync(processedImage);
+            BitmapImage outputImage = BitmapFunctions.ConvertBitmapToBitmapImage(processedImage);
 
             NarratorHelper.AnnounceImageChanged(DefaultImage, "Image changed: all regions segmented."); // <exclude-line>
 

AIDevGallery/Samples/Open Source Models/Image Models/Faster RCNN/ObjectDetection.xaml.cs

@@ -11,7 +11,6 @@
 using System;
 using System.Collections.Generic;
 using System.Drawing;
-using System.IO;
 using System.Linq;
 using System.Threading.Tasks;
 using Windows.Storage.Pickers;
@@ -31,7 +30,7 @@ namespace AIDevGallery.Samples.OpenSourceModels.ObjectDetection.FasterRCNN
             "Microsoft.ML.OnnxRuntime.DirectML",
             "Microsoft.ML.OnnxRuntime.Extensions"
         ],
-        Name = "Object Detection",
+        Name = "Faster RCNN Object Detection",
         Id = "9b74ccc0-f5f7-430f-bed0-758ffc063508",
         Icon = "\uE8B3")]
     internal sealed partial class ObjectDetection : BaseSamplePage
@@ -164,30 +163,18 @@ private async Task DetectObjects(string filePath)
                 return predictions;
             });
 
-            RenderPredictions(image, predictions);
-            image.Dispose();
-
-            Loader.IsActive = false;
-            Loader.Visibility = Visibility.Collapsed;
-            UploadButton.Visibility = Visibility.Visible;
-        }
-
-        private void RenderPredictions(Bitmap image, List<Prediction> predictions)
-        {
-            BitmapFunctions.DrawPredictions(image, predictions);
+            BitmapImage outputImage = BitmapFunctions.RenderPredictions(image, predictions);
 
-            BitmapImage bitmapImage = new();
-            using (MemoryStream memoryStream = new())
+            DispatcherQueue.TryEnqueue(() =>
             {
-                image.Save(memoryStream, System.Drawing.Imaging.ImageFormat.Png);
-
-                memoryStream.Position = 0;
-
-                bitmapImage.SetSource(memoryStream.AsRandomAccessStream());
-            }
+                DefaultImage.Source = outputImage;
+                Loader.IsActive = false;
+                Loader.Visibility = Visibility.Collapsed;
+                UploadButton.Visibility = Visibility.Visible;
+            });
 
-            DefaultImage.Source = bitmapImage;
             NarratorHelper.AnnounceImageChanged(DefaultImage, "Image changed: objects detected."); // <exclude-line>
+            image.Dispose();
         }
     }
 }

AIDevGallery/Samples/Open Source Models/Image Models/HRNetPose/PoseDetection.xaml.cs

@@ -24,7 +24,8 @@ namespace AIDevGallery.Samples.OpenSourceModels.HRNetPose
         SharedCode = [
             SharedCodeEnum.Prediction,
             SharedCodeEnum.BitmapFunctions,
-            SharedCodeEnum.DeviceUtils
+            SharedCodeEnum.DeviceUtils,
+            SharedCodeEnum.PoseHelper
         ],
         NugetPackageReferences = [
             "System.Drawing.Common",
@@ -145,13 +146,20 @@ private async Task DetectPose(string filePath)
                 // Run inference
                 using IDisposableReadOnlyCollection<DisposableNamedOnnxValue> results = _inferenceSession!.Run(inputs);
                 var heatmaps = results[0].AsTensor<float>();
-                List<(float X, float Y)> keypointCoordinates = PostProcessResults(heatmaps, originalImageWidth, originalImageHeight);
+
+                var outputName = _inferenceSession!.OutputNames[0];
+                var outputDimensions = _inferenceSession!.OutputMetadata[outputName].Dimensions;
+
+                float outputWidth = outputDimensions[2];
+                float outputHeight = outputDimensions[3];
+
+                List<(float X, float Y)> keypointCoordinates = PoseHelper.PostProcessResults(heatmaps, originalImageWidth, originalImageHeight, outputWidth, outputHeight);
                 return keypointCoordinates;
             });
 
             // Render predictions and create output bitmap
-            using Bitmap output = RenderPredictions(image, predictions);
-            BitmapImage outputImage = BitmapFunctions.ConvertBitmapToBitmapImageAsync(output);
+            using Bitmap output = PoseHelper.RenderPredictions(image, predictions, .02f);
+            BitmapImage outputImage = BitmapFunctions.ConvertBitmapToBitmapImage(output);
             NarratorHelper.AnnounceImageChanged(DefaultImage, "Image changed: key points rendered."); // <exclude-line>
 
             DispatcherQueue.TryEnqueue(() =>
@@ -162,88 +170,5 @@ private async Task DetectPose(string filePath)
                 UploadButton.Visibility = Visibility.Visible;
             });
         }
-
-        private List<(float X, float Y)> PostProcessResults(Tensor<float> heatmaps, float originalWidth, float originalHeight)
-        {
-            List<(float X, float Y)> keypointCoordinates = [];
-
-            // Scaling factors from heatmap (64x48) directly to original image size
-            float scale_x = originalWidth / 64f;
-            float scale_y = originalHeight / 48f;
-
-            int numKeypoints = heatmaps.Dimensions[1];
-            int heatmapWidth = heatmaps.Dimensions[2];
-            int heatmapHeight = heatmaps.Dimensions[3];
-
-            for (int i = 0; i < numKeypoints; i++)
-            {
-                float maxVal = float.MinValue;
-                int maxX = 0, maxY = 0;
-
-                for (int x = 0; x < heatmapWidth; x++)
-                {
-                    for (int y = 0; y < heatmapHeight; y++)
-                    {
-                        float value = heatmaps[0, i, y, x];
-                        if (value > maxVal)
-                        {
-                            maxVal = value;
-                            maxX = x;
-                            maxY = y;
-                        }
-                    }
-                }
-
-                float scaledX = maxX * scale_x;
-                float scaledY = maxY * scale_y;
-
-                keypointCoordinates.Add((scaledX, scaledY));
-            }
-
-            return keypointCoordinates;
-        }
-
-        private Bitmap RenderPredictions(Bitmap originalImage, List<(float X, float Y)> keypoints)
-        {
-            Bitmap outputImage = new(originalImage);
-
-            using (Graphics g = Graphics.FromImage(outputImage))
-            {
-                int markerSize = (int)((originalImage.Width + originalImage.Height) * 0.02 / 2);
-                Brush brush = Brushes.Red;
-
-                using Pen linePen = new(Color.Blue, 5);
-                List<(int StartIdx, int EndIdx)> connections =
-                [
-                    (5, 6),   // Left shoulder to right shoulder
-                        (5, 7),   // Left shoulder to left elbow
-                        (7, 9),   // Left elbow to left wrist
-                        (6, 8),   // Right shoulder to right elbow
-                        (8, 10),  // Right elbow to right wrist
-                        (11, 12), // Left hip to right hip
-                        (5, 11),  // Left shoulder to left hip
-                        (6, 12),  // Right shoulder to right hip
-                        (11, 13), // Left hip to left knee
-                        (13, 15), // Left knee to left ankle
-                        (12, 14), // Right hip to right knee
-                        (14, 16) // Right knee to right ankle
-                ];
-
-                foreach (var (startIdx, endIdx) in connections)
-                {
-                    var (startPointX, startPointY) = keypoints[startIdx];
-                    var (endPointX, endPointY) = keypoints[endIdx];
-
-                    g.DrawLine(linePen, startPointX, startPointY, endPointX, endPointY);
-                }
-
-                foreach (var (x, y) in keypoints)
-                {
-                    g.FillEllipse(brush, x - markerSize / 2, y - markerSize / 2, markerSize, markerSize);
-                }
-            }
-
-            return outputImage;
-        }
     }
 }

AIDevGallery/Samples/Open Source Models/Image Models/MultiHRNetPose/Multipose.xaml

@@ -0,0 +1,36 @@
+<?xml version="1.0" encoding="utf-8"?>
+<samples:BaseSamplePage
+    xmlns:samples="using:AIDevGallery.Samples"
+    x:Class="AIDevGallery.Samples.OpenSourceModels.MultiHRNetPose.Multipose"
+    xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
+    xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
+    xmlns:local="using:AIDevGallery.Samples.OpenSourceModels.MultiHRNetPose"
+    xmlns:d="http://schemas.microsoft.com/expression/blend/2008"
+    xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"
+    mc:Ignorable="d">
+
+  <ScrollViewer>
+    <Grid RowSpacing="16">
+      <Grid.RowDefinitions>
+        <RowDefinition Height="Auto" />
+        <RowDefinition Height="Auto" />
+      </Grid.RowDefinitions>
+      <Image
+            x:Name="DefaultImage"
+            MaxWidth="800"
+            MaxHeight="500" />
+      <ProgressRing
+            x:Name="Loader"
+            Grid.Row="1"
+            IsActive="false"
+            Visibility="Collapsed" />
+      <Button
+            x:Name="UploadButton"
+            Grid.Row="1"
+            HorizontalAlignment="Center"
+            Click="UploadButton_Click"
+            Content="Select image"
+            Style="{StaticResource AccentButtonStyle}" />
+    </Grid>
+  </ScrollViewer>
+</samples:BaseSamplePage>