Create a solution to demonstrate vector search in SQL Server with ML.NET

mcbethr · Nov 2, 2023 · c0f9483 · c0f9483
1 parent 5e63e80
commit c0f9483
Show file tree

Hide file tree

Showing 25 changed files with 646 additions and 0 deletions.
diff --git a/Database/Database.csproj b/Database/Database.csproj
@@ -0,0 +1,14 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+    <PropertyGroup>
+        <TargetFramework>net7.0</TargetFramework>
+        <ImplicitUsings>enable</ImplicitUsings>
+        <Nullable>enable</Nullable>
+    </PropertyGroup>
+
+    <ItemGroup>
+        <PackageReference Include="Microsoft.EntityFrameworkCore" Version="7.0.13"/>
+        <PackageReference Include="Microsoft.EntityFrameworkCore.SqlServer" Version="7.0.13"/>
+    </ItemGroup>
+
+</Project>
diff --git a/Database/Image.cs b/Database/Image.cs
@@ -0,0 +1,13 @@
+using System.ComponentModel.DataAnnotations;
+using System.ComponentModel.DataAnnotations.Schema;
+
+namespace Database;
+
+public class Image
+{
+    [Key]
+    [DatabaseGenerated(DatabaseGeneratedOption.Identity)]
+    public int Id { get; set; }
+
+    [Required] public string FilePath { get; set; }
+}
diff --git a/Database/ImageDbContext .cs b/Database/ImageDbContext .cs
@@ -0,0 +1,29 @@
+using Microsoft.EntityFrameworkCore;
+
+namespace Database;
+
+public class ImageDbContext : DbContext
+{
+    public DbSet<Image> Images { get; set; }
+    public DbSet<Vector> Vectors { get; set; }
+    public DbSet<SimilarImage> SimilarImages { get; set; }
+
+    public IQueryable<SimilarImage> GetSimilarImages(string vector)
+    {
+        return SimilarImages.FromSqlRaw(
+            "SELECT * FROM dbo.SimilarImages({0})", vector);
+    }
+
+    protected override void OnConfiguring(DbContextOptionsBuilder optionsBuilder)
+    {
+        optionsBuilder.UseSqlServer(
+            @"Server=localhost;Database=ReserveImageSearch;User=sa;Password=!MySecretTemplafyPassword1;TrustServerCertificate=true");
+    }
+
+    protected override void OnModelCreating(ModelBuilder modelBuilder)
+    {
+        modelBuilder.Entity<SimilarImage>().HasNoKey();
+        modelBuilder.Entity<Vector>().HasKey(v => new { v.ImageId, v.VectorPosition });
+        base.OnModelCreating(modelBuilder);
+    }
+}
diff --git a/Database/Scripts/0 - Create Database.sql b/Database/Scripts/0 - Create Database.sql
@@ -0,0 +1,2 @@
+CREATE
+DATABASE ReserveImageSearch
diff --git a/Database/Scripts/1 - Define Images Table.sql b/Database/Scripts/1 - Define Images Table.sql
@@ -0,0 +1,5 @@
+CREATE TABLE Images
+(
+    Id       INT PRIMARY KEY IDENTITY(1,1),
+    FilePath VARCHAR(255) NOT NULL
+);
diff --git a/Database/Scripts/2 - Define Vectors Table.sql b/Database/Scripts/2 - Define Vectors Table.sql
@@ -0,0 +1,7 @@
+CREATE TABLE Vectors
+(
+    ImageId        INT FOREIGN KEY REFERENCES Images(Id),
+    VectorPosition INT,
+    VectorValue    FLOAT,
+    PRIMARY KEY (ImageId, VectorPosition)
+);
diff --git a/Database/Scripts/3 - Define SimilarImages function.sql b/Database/Scripts/3 - Define SimilarImages function.sql
@@ -0,0 +1,36 @@
+CREATE
+OR
+ALTER FUNCTION dbo.SimilarImages(@vector NVARCHAR(max))
+    RETURNS TABLE
+    AS
+    RETURN WITH cteVector AS
+    (
+    SELECT
+    cast ([key] AS INT) AS [VectorPosition],
+    cast ([value] AS FLOAT) AS [VectorValue]
+    FROM
+    OPENJSON(@vector)
+    ),
+    cteSimilar AS
+    (
+    SELECT TOP (50)
+    v2.Id,
+    SUM (v1.[VectorValue] * v2.[VectorValue]) / (
+    SQRT(SUM (v1.[VectorValue] * v1.[VectorValue])) * SQRT(SUM (v2.[VectorValue] * v2.[VectorValue]))
+    ) as CosineDistance
+    FROM
+    cteVector v1
+    INNER JOIN
+    [dbo].[Vectors] v2 ON v1.VectorPosition = v2.VectorPosition
+    GROUP BY
+    v2.ImageId
+    ORDER BY
+    cosine_distance DESC
+    )
+SELECT i.FilePath,
+       r.CosineDistance
+FROM cteSimilar r
+         INNER JOIN
+    [dbo].[Images] i
+ON i.Id = r.ImageId
+    GO
diff --git a/Database/SimilarImage.cs b/Database/SimilarImage.cs
@@ -0,0 +1,7 @@
+namespace Database;
+
+public class SimilarImage
+{
+    public string FilePath { get; set; }
+    public double CosineDistance { get; set; }
+}
diff --git a/Database/Vector.cs b/Database/Vector.cs
@@ -0,0 +1,16 @@
+using System.ComponentModel.DataAnnotations.Schema;
+
+namespace Database;
+
+public class Vector
+{
+    [Column(Order = 0)]
+    [ForeignKey("Image")]
+    public int ImageId { get; set; }
+
+    [Column(Order = 1)] public int VectorPosition { get; set; }
+
+    public float VectorValue { get; set; }
+
+    public virtual Image Image { get; set; }
+}
diff --git a/DatasetBuilder/DatasetBuilder.csproj b/DatasetBuilder/DatasetBuilder.csproj
@@ -0,0 +1,216 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+    <PropertyGroup>
+        <OutputType>Exe</OutputType>
+        <TargetFramework>net7.0</TargetFramework>
+        <ImplicitUsings>enable</ImplicitUsings>
+        <Nullable>enable</Nullable>
+    </PropertyGroup>
+
+    <ItemGroup>
+        <ProjectReference Include="..\Database\Database.csproj"/>
+        <ProjectReference Include="..\Vectorizer\Vectorizer.csproj"/>
+    </ItemGroup>
+
+    <ItemGroup>
+        <None Update="Images\image1.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image10.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image11.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image12.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image13.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image14.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image15.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image16.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image17.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image18.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image19.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image2.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image20.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image21.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image22.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image23.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image24.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image25.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image26.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image27.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image28.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image29.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image3.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image30.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image31.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image32.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image33.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image34.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image35.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image36.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image37.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image38.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image39.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image4.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image40.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image41.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image42.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image43.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image44.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image45.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image46.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image47.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image48.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image49.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image5.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image50.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image51.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image52.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image53.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image54.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image55.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image56.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image57.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image58.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image59.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image6.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image60.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image61.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image62.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image63.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image64.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image65.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image66.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image7.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image8.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+        <None Update="Images\image9.jpg">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+    </ItemGroup>
+
+</Project>
diff --git a/DatasetBuilder/Images/image1.jpg b/DatasetBuilder/Images/image1.jpg