/*
* Process an image and identify what is in it. When done, the method
* {@link #onPhotoRecognitionReady(Collection)} must be called with the results of
* the image recognition process.
*
* @param image Bitmap containing the image to be classified. The image can be
* of any size, but preprocessing might occur to resize it to the
* format expected by the classification process, which can be time
* and power consuming.
*/
List <Recognition> DoRecognize(Bitmap image)
{
// Allocate space for the inference results
var count = mLabels.Count;
// Allocate buffer for image pixels.
int[] intValues = new int[TF_INPUT_IMAGE_WIDTH * TF_INPUT_IMAGE_HEIGHT];
ByteBuffer imgData = ByteBuffer.AllocateDirect(
4 * DIM_BATCH_SIZE * TF_INPUT_IMAGE_WIDTH * TF_INPUT_IMAGE_HEIGHT * DIM_PIXEL_SIZE);
imgData.Order(ByteOrder.NativeOrder());
// Read image data into buffer formatted for the TensorFlow model
TensorFlowHelper.ConvertBitmapToByteBuffer(image, intValues, imgData);
// Run inference on the network with the image bytes in imgData as input,
// storing results on the confidencePerLabel array. initialize arrays.
float[][] confidence = new float[1][];
confidence[0] = new float[count];
//wrap it inside a Java Object
var conf = FromArray <float[]>(confidence);
mTensorFlowLite.Run(imgData, conf);
//convert it back
confidence = conf.ToArray <float[]>();
List <Recognition> results = TensorFlowHelper.GetBestResults(confidence[0], mLabels);
return(results);
}
/*
* Process an image and identify what is in it. When done, the method
* {@link #onPhotoRecognitionReady(Collection)} must be called with the results of
* the image recognition process.
*
* @param image Bitmap containing the image to be classified. The image can be
* of any size, but preprocessing might occur to resize it to the
* format expected by the classification process, which can be time
* and power consuming.
*/
void DoRecognize(Bitmap image)
{
// Allocate space for the inference results
var count = mLabels.Count;
// Allocate buffer for image pixels.
int[] intValues = new int[TF_INPUT_IMAGE_WIDTH * TF_INPUT_IMAGE_HEIGHT];
//ByteBuffer imgData = ByteBuffer.AllocateDirect(
// 4 * DIM_BATCH_SIZE * TF_INPUT_IMAGE_WIDTH * TF_INPUT_IMAGE_HEIGHT * DIM_PIXEL_SIZE);
ByteBuffer imgData = ByteBuffer.AllocateDirect(
DIM_BATCH_SIZE * TF_INPUT_IMAGE_WIDTH * TF_INPUT_IMAGE_HEIGHT * DIM_PIXEL_SIZE);
imgData.Order(ByteOrder.NativeOrder());
// Read image data into buffer formatted for the TensorFlow model
TensorFlowHelper.ConvertBitmapToByteBuffer(image, intValues, imgData);
// Run inference on the network with the image bytes in imgData as input,
// storing results on the confidencePerLabel array.
//ByteBuffer confidenceByteBuffer = ByteBuffer.Allocate(count);
//mTensorFlowLite.Run(imgData, confidenceByteBuffer);
//byte[] confidenceByteArray = ConvertResults(confidenceByteBuffer);
//var confidenceBuffer = FloatBuffer.Allocate(4 * count);
byte[][] confidence = new byte[1][];
confidence[0] = new byte[count];
var conf = Java.Lang.Object.FromArray <byte[]>(confidence);
mTensorFlowLite.Run(imgData, conf);
confidence = conf.ToArray <byte[]>();
List <Recognition> results = TensorFlowHelper.GetBestResults(confidence[0], mLabels);
/*float[][] confidence = new float[1][];
* confidence[0] = new float[count];
*
* var conf = Java.Lang.Object.FromArray<float[]>(confidence);
* mTensorFlowLite.Run(imgData, conf);
*
* confidence = conf.ToArray<float[]>();
* List<Recognition> results = TensorFlowHelper.GetBestResults(confidence[0], mLabels);
*/
//float[] confidenceArray = ConvertResults(confidenceBuffer.AsFloatBuffer());
//float[] confidenceByteArray = ConvertResultsFloat(confidenceBuffer, count);
// Get the results with the highest confidence and map them to their labels
//List<Recognition> results = TensorFlowHelper.GetBestResults(confidenceArray, mLabels);
//List<Recognition> results = TensorFlowHelper.GetBestResults(confidenceByteArray, mLabels);
//List<Recognition> results = TensorFlowHelper.GetBestResults(confidencePerLabel, mLabels);
// Report the results with the highest confidence
OnPhotoRecognitionReady(results);
}
/*
* Initialize the classifier that will be used to process images.
*/
void InitClassifier()
{
try
{
mTensorFlowLite = new Interpreter(TensorFlowHelper.LoadModelFile(this, MODEL_FILE), 2);
mLabels = TensorFlowHelper.ReadLabels(this, LABELS_FILE);
}
catch (IOException e)
{
Log.Warn(TAG, "Unable to initialize TensorFlow Lite.", e);
}
}
