README.md

PerchEye iOS Integration Guide

This project demonstrates the integration and usage of PerchEye Framework for face recognition and comparison in iOS applications.

Requirements

• iOS 14.0+
• Xcode 12.0+
• Swift 5.0+

Library Integration

1. Adding PerchEye Framework

1. Download PerchEye Framework from the official source
2. Drag PerchEyeFramework.xcframework into your Xcode project
3. Ensure the framework is added to Frameworks, Libraries, and Embedded Content
4. Set Embed & Sign for the framework

2. Permissions Setup

Add photo library access permission to Info.plist:

<key>NSPhotoLibraryUsageDescription</key>
<string>App needs access to photos for face analysis</string>

Basic Usage

1. Import Library

import PerchEyeFramework

2. Create Face Analyzer

class FacesAnalyzer {
    private let perchEye: PerchEyeSwift
    
    init() {
        perchEye = PerchEyeSwift()
    }
}

3. Analyze Images

func analyzeFaces(in firstImages: [UIImage], and secondImages: [UIImage]) async throws -> Float {
    return try await withCheckedThrowingContinuation { continuation in
        do {
            // Open transaction for first image group
            perchEye.openTransaction()
            let firstHash = try generateHash(for: firstImages)
            
            // Open transaction for second image group
            perchEye.openTransaction()
            try loadImages(for: secondImages)
            
            // Compare images
            let similarity = perchEye.verify(hash: firstHash)
            
            continuation.resume(returning: similarity)
        } catch {
            continuation.resume(throwing: error)
        }
    }
}

4. Handle Results

func compareImages() async {
    do {
        let score = try await facesAnalyzer.analyzeFaces(in: firstImages, and: secondImages)
        
        // Compare with similarity threshold
        let threshold: Float = 0.8
        if score > threshold {
            print("Faces are similar: \(score)")
        } else {
            print("Faces are different: \(score)")
        }
    } catch {
        print("Analysis error: \(error)")
    }
}

Key Components

FacesAnalyzer

Main class for working with PerchEye SDK:

• analyzeFaces() - compare two groups of images
• generateHash() - generate hash for images
• loadImages() - load images for analysis

Error Handling

struct CustomError: Error, LocalizedError, Identifiable {
    let id = UUID()
    let title: String
    let message: String
}

Common errors:

• sdkNotInitialized - SDK not initialized
• faceNotFound - Face not found in image
• transactionNotOpened - Transaction not opened
• internalError - Internal SDK error

UI Components

PhotoFolderView

Component for photo selection:

PhotoFolderView(
    images: $viewModel.firstImagesGroup,
    selectedItems: $firstSelectedItems,
    isPresented: $showFirstPhotoPicker
)

Result Display

Text(String(format: "%.2f", score))
    .foregroundColor(score > threshold ? .green : .red)

SwiftUI Usage Example

struct ContentView: View {
    @StateObject var viewModel = ViewModel()
    
    var body: some View {
        VStack {
            // Photo selection UI
            HStack {
                PhotoFolderView(images: $viewModel.firstImagesGroup, ...)
                PhotoFolderView(images: $viewModel.secondImagesGroup, ...)
            }
            
            // Compare button
            Button("Compare Photos") {
                Task {
                    await viewModel.compareImages()
                }
            }
            
            // Result display
            if let score = viewModel.similarityScore {
                Text("Similarity: \(String(format: "%.2f", score))")
            }
        }
        .errorAlert($viewModel.error)
    }
}

Usage Tips

1. Image Quality: Use images with clearly visible faces
2. Lighting: Avoid images with poor lighting conditions
3. Face Count: Image should contain only one face
4. Similarity Threshold: Recommended threshold of 0.8 for accurate recognition
5. Error Handling: Always handle cases when face is not found

License

Check PerchEye Framework license terms before using in commercial projects.