TensorFlow Lite开发之——iOS图像分类识别(5)

发表于 分类于 ， ， 阅读时长 ≈ 3 分钟

一 概述

  • 图像模型介绍
  • iOS开发流程
  • 运行数据识别结果

二 图像模型介绍

2.1 模型位置

2.2 模型说明

  • 每个模型可以识别1000多种物体特征
  • 应用启动后，通过Model选项可以选择Model

三 iOS开发流程

3.1 将模型放到TFLite文件夹下

3.2 添加到Bundle Resources

3.3 在Podfile中添加TensorFlow Lite依赖

1
2
3
4
5
target 'ImageClassification' do
  # Comment the next line if you're not using Swift and don't want to use dynamic frameworks
  use_frameworks!
   pod 'TensorFlowLiteTaskVision'
end

3.4 界面布局

3.5 定义ImageClassificationHelper用于对图形解析

1
2
3
4
5
6
let options = ImageClassifierOptions(modelPath: modelPath)
options.baseOptions.computeSettings.cpuSettings.numThreads = Int(Int32(threadCount))
options.classificationOptions.maxResults = resultCount
options.classificationOptions.scoreThreshold = scoreThreshold

classifier = try ImageClassifier.classifier(options: options)

3.6 ViewController定义模型

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
enum ModelType: CaseIterable {
  case efficientnetLite0
  case efficientnetLite1
  case efficientnetLite2
  case efficientnetLite3
  case efficientnetLite4

  var modelFileInfo: FileInfo {
    switch self {
    case .efficientnetLite0:
      return FileInfo("efficientnet_lite0", "tflite")
    case .efficientnetLite1:
      return FileInfo("efficientnet_lite1", "tflite")
    case .efficientnetLite2:
      return FileInfo("efficientnet_lite2", "tflite")
    case .efficientnetLite3:
      return FileInfo("efficientnet_lite3", "tflite")
    case .efficientnetLite4:
      return FileInfo("efficientnet_lite4", "tflite")
    }
  }

  var title: String {
    switch self {
    case .efficientnetLite0:
      return "EfficientNet-Lite0"
    case .efficientnetLite1:
      return "EfficientNet-Lite1"
    case .efficientnetLite2:
      return "EfficientNet-Lite2"
    case .efficientnetLite3:
      return "EfficientNet-Lite3"
    case .efficientnetLite4:
      return "EfficientNet-Lite4"
    }
  }
}

3.7 识别结果

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
extension ViewController: CameraFeedManagerDelegate {

  func didOutput(pixelBuffer: CVPixelBuffer) {
    let currentTimeMs = Date().timeIntervalSince1970 * 1000
    guard (currentTimeMs - previousInferenceTimeMs) >= delayBetweenInferencesMs else { return }
    previousInferenceTimeMs = currentTimeMs
    guard !isInferenceQueueBusy else { return }
    inferenceQueue.async { [weak self] in
      guard let self = self else { return }

      self.isInferenceQueueBusy = true

      // Pass the pixel buffer to TensorFlow Lite to perform inference.
      let result = self.imageClassificationHelper?.classify(frame: pixelBuffer)

      self.isInferenceQueueBusy = false

      // Display results by handing off to the InferenceViewController.
      DispatchQueue.main.async {
        let resolution = CGSize(
          width: CVPixelBufferGetWidth(pixelBuffer), height: CVPixelBufferGetHeight(pixelBuffer))
        self.inferenceViewController?.inferenceResult = result
        self.inferenceViewController?.resolution = resolution
        self.inferenceViewController?.tableView.reloadData()
      }
    }
  }

四 运行数据识别结果

识别结果1 识别结果1 识别结果1

五 参考

  • TensorFlow-ios快速入门
  • TensorFlow Lite示例应用