TensorFlow的一个简单输出Demo

TensorFlow程序 简单输出程序

"""
 TensorFlow程序 简单输出程序
"""
from __future__ import print_function
# 导入tensorflow的包
import tensorflow as tf

# try except 
try:
    tf.contrib.eager.enable_eager_execution()
except ValueError:
    pass

# 输出结果
tensor = tf.constant('Gxs every day')
print('tensor:')
print(tensor)

print('输出:')
# 将tensor 转换成 numpy 形式输出
tensor_value = tensor.numpy()
print(tensor_value)

在这里插入图片描述

这里是一个基于TensorFlow Lite和Camera2 API的Android自动检测并拍照的Demo。 首先,在gradle文件中添加以下依赖: ``` implementation 'org.tensorflow:tensorflow-lite:2.5.0' implementation 'org.tensorflow:tensorflow-lite-gpu:2.5.0' ``` 接下来,创建一个Camera2的预览类CameraPreview,用于预览相机画面,并在其中初始化TensorFlow Lite模型。 ``` public class CameraPreview extends TextureView implements TextureView.SurfaceTextureListener, ImageReader.OnImageAvailableListener { private static final String MODEL_PATH = "model.tflite"; private static final String LABELS_PATH = "labels.txt"; private CameraDevice cameraDevice; private CameraCaptureSession cameraCaptureSession; private CaptureRequest.Builder captureRequestBuilder; private HandlerThread backgroundThread; private Handler backgroundHandler; private ImageReader imageReader; private Interpreter interpreter; private List<String> labels; public CameraPreview(Context context) { super(context); setSurfaceTextureListener(this); initModel(); } private void initModel() { try { // 加载模型 interpreter = new Interpreter(loadModelFile(), new Interpreter.Options()); // 加载标签 labels = loadLabelList(); } catch (IOException e) { e.printStackTrace(); } } private MappedByteBuffer loadModelFile() throws IOException { AssetFileDescriptor fileDescriptor = getContext().getAssets().openFd(MODEL_PATH); FileInputStream inputStream = new FileInputStream(fileDescriptor.getFileDescriptor()); FileChannel fileChannel = inputStream.getChannel(); long startOffset = fileDescriptor.getStartOffset(); long declaredLength = fileDescriptor.getDeclaredLength(); return fileChannel.map(FileChannel.MapMode.READ_ONLY, startOffset, declaredLength); } private List<String> loadLabelList() throws IOException { List<String> labelList = new ArrayList<>(); BufferedReader reader = new BufferedReader(new InputStreamReader(getContext().getAssets().open(LABELS_PATH))); String line; while ((line = reader.readLine()) != null) { labelList.add(line); } reader.close(); return labelList; } @Override public void onSurfaceTextureAvailable(SurfaceTexture surfaceTexture, int width, int height) { openCamera(); } private void openCamera() { CameraManager cameraManager = (CameraManager) getContext().getSystemService(Context.CAMERA_SERVICE); try { String cameraId = cameraManager.getCameraIdList()[0]; CameraCharacteristics cameraCharacteristics = cameraManager.getCameraCharacteristics(cameraId); Size[] outputSizes = cameraCharacteristics.get(CameraCharacteristics.SCALER_STREAM_CONFIGURATION_MAP).getOutputSizes(ImageFormat.JPEG); imageReader = ImageReader.newInstance(outputSizes[0].getWidth(), outputSizes[0].getHeight(), ImageFormat.JPEG, 1); imageReader.setOnImageAvailableListener(this, backgroundHandler); if (ActivityCompat.checkSelfPermission(getContext(), Manifest.permission.CAMERA) != PackageManager.PERMISSION_GRANTED) { return; } cameraManager.openCamera(cameraId, new CameraDevice.StateCallback() { @Override public void onOpened(CameraDevice cameraDevice) { CameraPreview.this.cameraDevice = cameraDevice; createCameraPreviewSession(); } @Override public void onDisconnected(CameraDevice cameraDevice) { cameraDevice.close(); CameraPreview.this.cameraDevice = null; } @Override public void onError(CameraDevice cameraDevice, int error) { cameraDevice.close(); CameraPreview.this.cameraDevice = null; } }, backgroundHandler); } catch (CameraAccessException e) { e.printStackTrace(); } } private void createCameraPreviewSession() { SurfaceTexture surfaceTexture = getSurfaceTexture(); surfaceTexture.setDefaultBufferSize(1920, 1080); Surface previewSurface = new Surface(surfaceTexture); Surface readerSurface = imageReader.getSurface(); try { captureRequestBuilder = cameraDevice.createCaptureRequest(CameraDevice.TEMPLATE_PREVIEW); captureRequestBuilder.addTarget(previewSurface); captureRequestBuilder.addTarget(readerSurface); cameraDevice.createCaptureSession(Arrays.asList(previewSurface, readerSurface), new CameraCaptureSession.StateCallback() { @Override public void onConfigured(CameraCaptureSession cameraCaptureSession) { CameraPreview.this.cameraCaptureSession = cameraCaptureSession; updatePreview(); } @Override public void onConfigureFailed(CameraCaptureSession cameraCaptureSession) { } }, backgroundHandler); } catch (CameraAccessException e) { e.printStackTrace(); } } private void updatePreview() { if (cameraDevice == null) { return; } captureRequestBuilder.set(CaptureRequest.CONTROL_MODE, CameraMetadata.CONTROL_MODE_AUTO); try { cameraCaptureSession.setRepeatingRequest(captureRequestBuilder.build(), null, backgroundHandler); } catch (CameraAccessException e) { e.printStackTrace(); } } @Override public void onSurfaceTextureSizeChanged(SurfaceTexture surfaceTexture, int width, int height) { } @Override public boolean onSurfaceTextureDestroyed(SurfaceTexture surfaceTexture) { closeCamera(); return true; } private void closeCamera() { if (cameraDevice != null) { cameraDevice.close(); cameraDevice = null; } if (cameraCaptureSession != null) { cameraCaptureSession.close(); cameraCaptureSession = null; } if (imageReader != null) { imageReader.close(); imageReader = null; } } @Override public void onSurfaceTextureUpdated(SurfaceTexture surfaceTexture) { } @Override public void onImageAvailable(ImageReader reader) { Image image = reader.acquireLatestImage(); Bitmap bitmap = getBitmap(image); image.close(); // 在子线程中进行模型推理 new Thread(() -> { String result = recognize(bitmap); if (result.equals("cat")) { // 拍照 takePicture(); } }).start(); } private Bitmap getBitmap(Image image) { ByteBuffer buffer = image.getPlanes()[0].getBuffer(); byte[] bytes = new byte[buffer.remaining()]; buffer.get(bytes); return BitmapFactory.decodeByteArray(bytes, 0, bytes.length); } private String recognize(Bitmap bitmap) { Bitmap resizedBitmap = Bitmap.createScaledBitmap(bitmap, 224, 224, true); TensorImage inputImage = new TensorImage(DataType.FLOAT32); inputImage.load(resizedBitmap); TensorBuffer outputBuffer = TensorBuffer.createFixedSize(new int[]{1, labels.size()}, DataType.FLOAT32); interpreter.run(inputImage.getBuffer(), outputBuffer.getBuffer()); float[] results = outputBuffer.getFloatArray(); int index = getMaxIndex(results); return labels.get(index); } private int getMaxIndex(float[] array) { int maxIndex = 0; float max = array[maxIndex]; for (int i = 1; i < array.length; i++) { if (array[i] > max) { max = array[i]; maxIndex = i; } } return maxIndex; } private void takePicture() { try { CaptureRequest.Builder builder = cameraDevice.createCaptureRequest(CameraDevice.TEMPLATE_STILL_CAPTURE); builder.addTarget(imageReader.getSurface()); builder.set(CaptureRequest.CONTROL_MODE, CameraMetadata.CONTROL_MODE_AUTO); builder.set(CaptureRequest.JPEG_ORIENTATION, getOrientation()); cameraCaptureSession.stopRepeating(); cameraCaptureSession.abortCaptures(); cameraCaptureSession.capture(builder.build(), null, null); } catch (CameraAccessException e) { e.printStackTrace(); } } private int getOrientation() { int rotation = ((Activity) getContext()).getWindowManager().getDefaultDisplay().getRotation(); int sensorOrientation = cameraDevice.getCameraCharacteristics(CameraCharacteristics.SENSOR_ORIENTATION); return (rotation + sensorOrientation + 270) % 360; } public void startBackgroundThread() { backgroundThread = new HandlerThread("Camera Background"); backgroundThread.start(); backgroundHandler = new Handler(backgroundThread.getLooper()); } public void stopBackgroundThread() { backgroundThread.quitSafely(); try { backgroundThread.join(); backgroundThread = null; backgroundHandler = null; } catch (InterruptedException e) { e.printStackTrace(); } } } ``` 然后,在Activity中使用CameraPreview类,并在onResume和onPause方法中开始和停止后台线程。 ``` public class MainActivity extends AppCompatActivity { private CameraPreview cameraPreview; @Override protected void onCreate(Bundle savedInstanceState) { super.onCreate(savedInstanceState); setContentView(R.layout.activity_main); cameraPreview = new CameraPreview(this); FrameLayout previewLayout = findViewById(R.id.preview_layout); previewLayout.addView(cameraPreview); } @Override protected void onResume() { super.onResume(); cameraPreview.startBackgroundThread(); } @Override protected void onPause() { cameraPreview.stopBackgroundThread(); super.onPause(); } } ``` 注意,这个Demo还需要一张名为"labels.txt"的标签文件和一个名为"model.tflite"的TensorFlow Lite模型文件,应该将它们放在assets目录下。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

北顾丶

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付:¥1
获取中
扫码支付

您的余额不足,请更换扫码支付或充值

打赏作者

实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值