Jetpack Compose中CameraX绑定ImageAnalysis时出现Not bound to a valid camera错误

问题现象 #

• 使用CameraX图像分析功能检测人脸微笑、眨眼动作时，执行cameraProvider.bindToLifecycle(lifecycleOwner, cameraSelector, preview, imageAnalysis)会触发「Not Bound To A valid camera: ImageCapture CameraX」错误
• 将绑定对象替换为preview和imageCapture时，代码可正常运行

问题原因 #

1. ImageAnalysis分辨率设置错误：代码中直接使用未完成布局的previewView.width和previewView.height作为目标分辨率，此时宽高值为0，CameraX无法匹配到相机支持的输出规格，导致绑定失败。
2. 用例绑定不完整：创建了imageCapture实例但未将其绑定到生命周期，后续调用takePhoto时，未绑定的imageCapture会触发相机有效性校验错误。

解决方案 #

1. 移除错误的分辨率设置：让CameraX自动适配相机支持的分辨率（推荐），或在previewView布局完成后获取真实宽高再设置。
2. 统一绑定所有用例：将imageCapture加入bindToLifecycle的参数列表，确保所有相机用例都绑定到生命周期。

修改后的完整代码 #

@Composable
fun FaceDetectorCameraView(
    hasPassedAllChecks: Boolean,
    onImageCaptured: (Uri) -> Unit,
    onErrorFace: (String) -> Unit,
    executor: Executor,
    doProcessFace: (List<Face>) -> Unit,
    onError: (ImageCaptureException) -> Unit 
) {

    val cameraSelector = CameraSelector.DEFAULT_FRONT_CAMERA
    val context = LocalContext.current
    val lifecycleOwner = LocalLifecycleOwner.current

    val preview = Preview.Builder().build()
    val previewView = remember { PreviewView(context) }
    val imageCapture: ImageCapture = remember { ImageCapture.Builder().build() }
    val imageAnalysis = ImageAnalysis.Builder()
        .setBackpressureStrategy(ImageAnalysis.STRATEGY_KEEP_ONLY_LATEST)
        .setImageQueueDepth(10)
        .build()
        .apply {
            setAnalyzer(executor, FaceAnalyzer(
                object : FaceAnalyzerCallback{
                    override fun processFace(faces: List<Face>) {
                        doProcessFace(faces)
                    }
                    override fun errorFace(error: String) {
                        onErrorFace(error)
                    }
                }
            ))
        }

    LaunchedEffect(cameraSelector) {
        val cameraProvider = context.getCameraProvider()
        // 绑定所有用到的相机用例
        cameraProvider.bindToLifecycle(
            lifecycleOwner, cameraSelector, preview, imageAnalysis, imageCapture
        )
        preview.setSurfaceProvider(previewView.surfaceProvider)
    }

    AndroidView(
        modifier = Modifier
            .fillMaxWidth()
            .requiredHeightIn(80.dp)
            .padding(start = 20.dp, end = 20.dp, top = 10.dp),
        factory = { ctx ->
            previewView
        },
    )

    if(hasPassedAllChecks){
        LaunchedEffect(true){
            delay(1500)
            takePhoto(
                contentResolver = context.contentResolver,
                imageCapture = imageCapture,
                executor = executor,
                onImageCaptured = onImageCaptured,
                onError = onError
            )
        }
    }
}

函数调用方式保持不变：

FaceDetectorCameraView(
    hasPassedAllChecks = hasPassedAllChecks,
    onImageCaptured = {
        onImageTaken(it)
    },
    onErrorFace = {},
    executor = ContextCompat.getMainExecutor(context),
    doProcessFace = doProcessFace,
    onError = {
         Toast.makeText(context, it.message, Toast.LENGTH_SHORT).show()
    },
)

可选：指定真实分辨率的实现 #

如果必须自定义目标分辨率，可通过监听previewView布局完成事件获取真实宽高：

LaunchedEffect(previewView) {
    previewView.viewTreeObserver.addOnGlobalLayoutListener(object : ViewTreeObserver.OnGlobalLayoutListener {
        override fun onGlobalLayout() {
            previewView.viewTreeObserver.removeOnGlobalLayoutListener(this)
            val realSize = Size(previewView.width, previewView.height)
            val updatedImageAnalysis = ImageAnalysis.Builder()
                .setTargetResolution(realSize)
                .setBackpressureStrategy(ImageAnalysis.STRATEGY_KEEP_ONLY_LATEST)
                .setImageQueueDepth(10)
                .build()
                .apply {
                    setAnalyzer(executor, FaceAnalyzer(
                        object : FaceAnalyzerCallback{
                            override fun processFace(faces: List<Face>) {
                                doProcessFace(faces)
                            }
                            override fun errorFace(error: String) {
                                onErrorFace(error)
                            }
                        }
                    ))
                }
            // 重新绑定用例
            val cameraProvider = context.getCameraProvider()
            cameraProvider.unbindAll()
            cameraProvider.bindToLifecycle(lifecycleOwner, cameraSelector, preview, updatedImageAnalysis, imageCapture)
        }
    })
}

内容的提问来源于stack exchange，提问作者ibramazin