Question

我正在查看使用视觉检测文本的应用程序中的内存泄漏。

我遇到了内存泄漏，当使用树时，该泄漏指向此行：

try imageRequestHandler.perform([self.textDetectionRequest])

我不确定为什么，希望有人可以提供帮助。

下面的完整代码。

private func performVisionRequest(image: CGImage, orientation: CGImagePropertyOrientation) {

    DispatchQueue.global(qos: .userInitiated).async {
        do {
            var imageRequestHandler = VNImageRequestHandler(cgImage: image, orientation: orientation, options: [:])
            try imageRequestHandler.perform([self.textDetectionRequest])
        } catch let error as NSError {
            print("Failed to perform vision request: \(error)")

        }
    }
}

这是整个班级：

import UIKit
import Vision

var noText: Bool!
var imageNo: UIImage!

internal class Slicer {

private var image = UIImage()
private var sliceCompletion: ((_ slices: [UIImage]) -> Void) = { _ in }

private lazy var textDetectionRequest: VNDetectTextRectanglesRequest = {

    return VNDetectTextRectanglesRequest(completionHandler: self.handleDetectedText)
}()

internal func slice(image: UIImage, completion: @escaping ((_: [UIImage]) -> Void)) {
    self.image = image
    self.sliceCompletion = completion
    self.performVisionRequest(image: image.cgImage!, orientation: .up)
}

// MARK: - Vision

private func performVisionRequest(image: CGImage, orientation: CGImagePropertyOrientation) {

    DispatchQueue.global(qos: .userInitiated).async {
        do {
            let imageRequestHandler = VNImageRequestHandler(cgImage: image, orientation: orientation, options: [:])
            try imageRequestHandler.perform([self.textDetectionRequest])
        } catch let error as NSError {
            self.sliceCompletion([UIImage]())
            print("Failed to perform vision request: \(error)")

        }
    }
}

private func handleDetectedText(request: VNRequest?, error: Error?) {
    if let err = error as NSError? {
        print("Failed during detection: \(err.localizedDescription)")
        return
    }
    guard let results = request?.results as? [VNTextObservation], !results.isEmpty else {

        noText = true
        print("Tony no text found")
        var slices = [imageNo]
        self.sliceCompletion(slices as! [UIImage])
        slices = []
        return }

    noText = false
    self.sliceImage(text: results, onImageWithBounds: CGRect(x: 0, y: 0, width: self.image.cgImage!.width, height: self.image.cgImage!.height))
}

private func sliceImage(text: [VNTextObservation], onImageWithBounds bounds: CGRect) {
    CATransaction.begin()

    var slices = [UIImage]()

    for wordObservation in text {
        let wordBox = boundingBox(forRegionOfInterest: wordObservation.boundingBox, withinImageBounds: bounds)

        if !wordBox.isNull {
            guard let slice = self.image.cgImage?.cropping(to: wordBox) else { continue }
            slices.append(UIImage(cgImage: slice))
        }
    }

    self.sliceCompletion(slices)

    CATransaction.commit()
}

private func boundingBox(forRegionOfInterest: CGRect, withinImageBounds bounds: CGRect) -> CGRect {

    let imageWidth = bounds.width
    let imageHeight = bounds.height

    // Begin with input rect.
    var rect = forRegionOfInterest

    // Reposition origin.
    rect.origin.x *= imageWidth
    rect.origin.y = ((1 - rect.origin.y) * imageHeight) - (forRegionOfInterest.height * imageHeight)

    // Rescale normalized coordinates. Tony adde + 30 to increase the size of rect
    rect.size.width *= imageWidth + 30
    rect.size.height *= imageHeight + 30

    return rect
}
}

Answer 1

其他人都告诉你的是正确的。您有两个引用SearchProfile（Slicer）实例的闭包，并且您需要打破这两个实例的保留周期。我认为这行是一个巨大的错误：

self

除保留周期外，您一无所获。删除这些行！相反，只需在需要时创建匿名函数。替换为：

private lazy var textDetectionRequest: VNDetectTextRectanglesRequest = {
    return VNDetectTextRectanglesRequest(completionHandler: self.handleDetectedText)
}()

与此：

try imageRequestHandler.perform([self.textDetectionRequest])

如果仍然有泄漏（我怀疑），则将其更改为

try imageRequestHandler.perform(
    [VNDetectTextRectanglesRequest(completionHandler:{ req, err in
        self.handleDetectedText(request:req, error:err)
    })]
)

Answer 2

在textDetectionRequest闭包中使用capture list：

lazy var textDetectionRequest: VNDetectTextRectanglesRequest = 
         { [weak self] in
             return VNDetectTextRectanglesRequest(completionHandler: self?.handleDetectedText)
         }()

Answer 3

我认为应该是这样的：

DispatchQueue.global(qos: .userInitiated).async { [weak self] _ in
    do {
        var imageRequestHandler = VNImageRequestHandler(cgImage: image, orientation: orientation, options: [:])
        try imageRequestHandler.perform([self?.textDetectionRequest])
    } catch let error as Error {
        print("Failed to perform vision request: \(error)")

    }
}

Answer 4

self.textDetectionRequest是lazy var，它将捕获handleDetectedText，而closure是[unowned self]，因此您应使用[weak self]或private lazy var textDetectionRequest: VNDetectTextRectanglesRequest = { [unowned self] in return VNDetectTextRectanglesRequest(completionHandler: self.handleDetectedText) }()之类的捕获列表避免保留周期。

VNRequest

另外，您可以将performVisionRequest数组作为依赖项传递给internal func slice(image: UIImage, completion: @escaping ((_: [UIImage]) -> Void)) { self.image = image self.sliceCompletion = completion let requests = self.textDetectionRequest self.performVisionRequest(image: image.cgImage!, orientation: .up, requests: [requests]) } // MARK: - Vision private func performVisionRequest(image: CGImage, orientation: CGImagePropertyOrientation, requests: [VNRequest]) { DispatchQueue.global(qos: .userInitiated).async { do { let imageRequestHandler = VNImageRequestHandler(cgImage: image, orientation: orientation, options: [:]) try imageRequestHandler.perform(requests) } catch let error as NSError { self.sliceCompletion([UIImage]()) print("Failed to perform vision request: \(error)") } } }函数，以简化代码

name

内存泄漏，在do-catch块中。 iOS，Swift

4 个答案: