Level : WORDPRESS BOOK LINKEDIN PATENT Send Mail 동냥하기 hajunho.com

반응형

https://www.google.com/search?q=%22Failed+to+invoke+the+interpreter+with+error%22&oq=%22Failed+to+invoke+the+interpreter+with+error%22&aqs=chrome.0.69i59j69i60.4768j0j7&sourceid=chrome&ie=UTF-8

https://books.google.co.kr/books?id=e--oDwAAQBAJ&pg=PA145&lpg=PA145&dq=%22Failed+to+invoke+the+interpreter+with+error%22&source=bl&ots=CqkTVIC9Gb&sig=ACfU3U1g0P6zHckN-f8lYSDoypASxlZMrw&hl=ko&sa=X&ved=2ahUKEwi0wpGpv7vmAhUBMN4KHWDdAG4Q6AEwAHoECAkQAQ#v=onepage&q=%22Failed%20to%20invoke%20the%20interpreter%20with%20error%22&f=false


func runModel(onFrame pixelBuffer: CVPixelBuffer) -> Result? {
    let sourcePixelFormat = CVPixelBufferGetPixelFormatType(pixelBuffer)
    assert(sourcePixelFormat == kCVPixelFormatType_32ARGB ||
             sourcePixelFormat == kCVPixelFormatType_32BGRA ||
               sourcePixelFormat == kCVPixelFormatType_32RGBA)


    let imageChannels = 4
    assert(imageChannels >= inputChannels)

    // Crops the image to the biggest square in the center and scales it down to model dimensions.
    let scaledSize = CGSize(width: inputWidth, height: inputHeight)
    guard let thumbnailPixelBuffer = pixelBuffer.centerThumbnail(ofSize: scaledSize) else {
      return nil
    }

    let interval: TimeInterval
    let outputTensor: Tensor
    do {
      let inputTensor = try interpreter.input(at: 0)

      // Remove the alpha component from the image buffer to get the RGB data.
      guard let rgbData = rgbDataFromBuffer(
        thumbnailPixelBuffer,
        byteCount: batchSize * inputWidth * inputHeight * inputChannels,
        isModelQuantized: inputTensor.dataType == .uInt8
      ) else {
        print("Failed to convert the image buffer to RGB data.")
        return nil
      }

      // Copy the RGB data to the input `Tensor`.
      try interpreter.copy(rgbData, toInputAt: 0)

      // Run inference by invoking the `Interpreter`.
      let startDate = Date()
      try interpreter.invoke()
      interval = Date().timeIntervalSince(startDate) * 1000

      // Get the output `Tensor` to process the inference results.
      outputTensor = try interpreter.output(at: 0)
    } catch let error {
      print("Failed to invoke the interpreter with error: \(error.localizedDescription)")
      return nil
    }

    let results: [Float]
    switch outputTensor.dataType {
    case .uInt8:
      guard let quantization = outputTensor.quantizationParameters else {
        print("No results returned because the quantization values for the output tensor are nil.")
        return nil
      }
      let quantizedResults = [UInt8](outputTensor.data)
      results = quantizedResults.map {
        quantization.scale * Float(Int($0) - quantization.zeroPoint)
      }
    case .float32:
      results = [Float32](unsafeData: outputTensor.data) ?? []
    default:
      print("Output tensor data type \(outputTensor.dataType) is unsupported for this example app.")
      return nil
    }

    // Process the results.
    let topNInferences = getTopN(results: results)

    // Return the inference time and inference results.
    return Result(inferenceTime: interval, inferences: topNInferences)
  }


Xcode 11
/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Library/Developer/CoreSimulator/Profiles/Runtimes/iOS.simruntime/Contents/Resources/RuntimeRoot/usr/lib/
put tbd copy to the path:

/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator.sdk/usr/lib/

architecture x86_64 in file 


I've faced the same issue after updating to xcode 10. I've resolved this issue by downloading libstdc++.6.0.9.tbd and libstdc++.6.tbd from https://github.com/Kila2/libstdc-.6.0.9.tbd.

then i copied these files to my project folder. After that I added these libs in Linked Frameworks and Libraries by following these steps: Project > Target > General > Linked Frameworks and Libraries click (+) > Click 'Add Other' > Navigat to your project folder in 'Open File' and select these libraries and add.

Now run the project it will run perfectly..


변환한 텐서 모델은 안드로이드에서 잘 되는데 iOS 지원 바로 안되게 되어 있는 듯.
libc 6.0은 dylib가 아니라 tbd만 올려서 해결. 시뮬레이터에서는 안됨. 제길슨.



반응형

'살며 생각하며,' 카테고리의 다른 글

아이폰 개발자가 추천하는 캘린더  (0) 2019.12.23
063  (0) 2019.12.19
개 발 일 지 061  (0) 2019.12.17
개 발 일 지 060  (0) 2019.12.13
개 발 일 지 059  (0) 2019.12.13
  • 네이버 블러그 공유하기
  • 네이버 밴드에 공유하기
  • 페이스북 공유하기
  • 카카오스토리 공유하기