Xcode集成CoreML图像模型报错：No exact matches in call to instance method 'prediction'

这个错误的核心原因是：你传入的SwiftUI Image类型和CoreML模型DDAMFN的prediction方法要求的输入类型不兼容。SwiftUI的Image是视图组件，不是原始图像数据，CoreML无法直接处理它。

解决步骤及代码示例 #

1. 先明确模型输入类型

打开你的.mlmodel文件，查看输入参数的类型——图像类CoreML模型通常要求CVPixelBuffer或CGImage格式，而非SwiftUI的Image。

2. 添加图像格式转换工具

需要把SwiftUI Image转换成CoreML支持的CVPixelBuffer，以下是封装好的转换扩展：

import UIKit
import SwiftUI
import CoreML

// SwiftUI Image转UIImage
extension Image {
    var uiImage: UIImage? {
        let hostingController = UIHostingController(rootView: self)
        hostingController.view.frame = CGRect(x: 0, y: 0, width: 1, height: 1)
        return hostingController.view.asImage()
    }
}

// UIView转UIImage
extension UIView {
    func asImage() -> UIImage {
        let renderer = UIGraphicsImageRenderer(bounds: bounds)
        return renderer.image { context in
            layer.render(in: context.cgContext)
        }
    }
}

// UIImage转CVPixelBuffer（适配模型要求的112×112尺寸）
extension UIImage {
    func toPixelBuffer(targetSize: CGSize) -> CVPixelBuffer? {
        let attributes = [
            kCVPixelBufferCGImageCompatibilityKey: kCFBooleanTrue,
            kCVPixelBufferCGBitmapContextCompatibilityKey: kCFBooleanTrue
        ] as CFDictionary
        
        var pixelBuffer: CVPixelBuffer?
        let status = CVPixelBufferCreate(
            kCFAllocatorDefault,
            Int(targetSize.width),
            Int(targetSize.height),
            kCVPixelFormatType_32ARGB,
            attributes,
            &pixelBuffer
        )
        
        guard status == kCVReturnSuccess, let buffer = pixelBuffer else {
            return nil
        }
        
        CVPixelBufferLockBaseAddress(buffer, [])
        let pixelData = CVPixelBufferGetBaseAddress(buffer)
        
        let rgbSpace = CGColorSpaceCreateDeviceRGB()
        guard let context = CGContext(
            data: pixelData,
            width: Int(targetSize.width),
            height: Int(targetSize.height),
            bitsPerComponent: 8,
            bytesPerRow: CVPixelBufferGetBytesPerRow(buffer),
            space: rgbSpace,
            bitmapInfo: CGImageAlphaInfo.noneSkipFirst.rawValue
        ) else {
            return nil
        }
        
        context.draw(self.cgImage!, in: CGRect(origin: .zero, size: targetSize))
        CVPixelBufferUnlockBaseAddress(buffer, [])
        
        return buffer
    }
}

3. 修改预测函数，传入正确格式的图像

let testImage: Image = Image("testHappyImage")

func testing(image: Image) -> DDAMFNOutput? {
    do {
        let config = MLModelConfiguration()
        let model = try DDAMFN(configuration: config)
        
        // 转换为模型要求的112×112尺寸的CVPixelBuffer
        guard let pixelBuffer = image.uiImage?.toPixelBuffer(targetSize: CGSize(width: 112, height: 112)) else {
            return nil
        }
        
        // 注意：参数名要和.mlmodel中定义的输入参数名完全一致（比如模型输入叫"image"就用image:pixelBuffer）
        let prediction = try model.prediction(image: pixelBuffer)
        return prediction
        
    } catch {
        print("预测失败：\(error)")
    }
    return nil
}

关键注意点 #

• 必须保证转换后的图像尺寸和模型要求的112×112完全一致，否则会触发尺寸不匹配的错误。
• 调用prediction时的参数名要和.mlmodel里输入参数的名称严格对应，比如模型输入参数名为inputImage，就要改成prediction(inputImage: pixelBuffer)。

内容的提问来源于stack exchange，提问作者BriteKnight