2

我在一个反应​​原生项目中使用 MLKit 和 iOS。

基本上使用此代码：https ://firebase.google.com/docs/ml/ios/label-images-with-automl

它曾经工作正常，但现在我收到此错误：

downloadModel: notificationDidFail: 
name = com.google.mlkit.notif.model-download-did-fail, 
object = Optional(<MLKModelDownloader: 0x281a86300>), 
userInfo = Optional([AnyHashable("MLKModelDownloadUserInfoKeyError"): Error Domain=com.google.mlkit Code=2 "Failed to save AutoML remote model labels file." UserInfo={NSLocalizedDescription=Failed to save AutoML remote model labels file.}, AnyHashable("MLKModelDownloadUserInfoKeyRemoteModel"): name: lepidoptera_underside])

Xcode 打印： [MLKit][I-MLK018012] AutoML 远程模型推理信息不包含标签

一种模型似乎比其他模型更有效。

这被执行：

NotificationCenter.default.addObserver(
  forName: .mlkitModelDownloadDidFail,
  object: nil,
  queue: nil
) { notification in
  self.failed("downloadModel: notification error: \(notification)")
}

整个下载模型功能：

func downloadModel() {
    guard let remoteModel = remoteModel else {
      failed("downloadModel: remoteModel error")
      return
    }

    let downloadConditions = ModelDownloadConditions(
      allowsCellularAccess: true,
      allowsBackgroundDownloading: true
    )

    _ = ModelManager.modelManager().download(
      remoteModel,
      conditions: downloadConditions
    )

    NotificationCenter.default.addObserver(
      forName: .mlkitModelDownloadDidSucceed,
      object: nil,
      queue: nil
    ) { [weak self] notification in
      guard let strongSelf = self,
            let userInfo = notification.userInfo,
            let model = userInfo[ModelDownloadUserInfoKey.remoteModel.rawValue]
              as? RemoteModel,
            model.name == strongSelf.modelName
      else { return }
      strongSelf.createLabeler()
    }

    NotificationCenter.default.addObserver(
      forName: .mlkitModelDownloadDidFail,
      object: nil,
      queue: nil
    ) { notification in
      self.failed("downloadModel: notification error: \(notification)")
    }
  }
4

1 回答 1

2

提供的文档 ( https://firebase.google.com/docs/ml/ios/label-images-with-automl ) 包含过时的信息。ML Kit 在其最新版本中已完全弃用并删除了该GoogleMLKit/ImageLabelingAutoMLpod。该吊舱现在被GoogleMLKit/ImageLabelingCustom吊舱取代。最新版本的MLKitImageLabelingCustompod 是1.2.0. 请参阅此处的完整迁移指南：

https://developers.google.com/ml-kit/vision/image-labeling/automl/migrate-automl

于 2021-03-12T19:40:50.437 回答