org.deeplearning4j.nn.layers.objdetect

Class DetectedObject

java.lang.Object

org.deeplearning4j.nn.layers.objdetect.DetectedObject

public class DetectedObject
extends Object

A detected object, by an object detection algorithm. Note that the dimensions (for center X/Y, width/height) depend on the specific implementation. For example, in the Yolo2OutputLayer, the dimensions are grid cell units - for example, with 416x416 input, 32x downsampling, we have 13x13 grid cells (each corresponding to 32 pixels in the input image). Thus, a centerX of 5.5 would be xPixels=5.5x32 = 176 pixels from left. Widths and heights are similar: in this example, a with of 13 would be the entire image (416 pixels), and a height of 6.5 would be 6.5/13 = 0.5 of the image (208 pixels).

Author:

Alex Black

Constructor Summary

Constructors

Constructor and Description
DetectedObject(int exampleNumber, double centerX, double centerY, double width, double height, org.nd4j.linalg.api.ndarray.INDArray classPredictions, double confidence)

Method Summary

Modifier and Type	Method and Description
double[]	getBottomRightXY() Get the bottom right X/Y coordinates of the detected object
int	getPredictedClass() Get the index of the predicted class (based on maximum predicted probability)
double[]	getTopLeftXY() Get the top left X/Y coordinates of the detected object
String	toString()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

DetectedObject

public DetectedObject(int exampleNumber,
                      double centerX,
                      double centerY,
                      double width,
                      double height,
                      org.nd4j.linalg.api.ndarray.INDArray classPredictions,
                      double confidence)

Parameters:

exampleNumber - Index of the example in the current minibatch. For single images, this is always 0

centerX - Center X position of the detected object

centerY - Center Y position of the detected object

width - Width of the detected object

height - Height of the detected object

classPredictions - Row vector of class probabilities for the detected object

Method Detail

getTopLeftXY

public double[] getTopLeftXY()

Get the top left X/Y coordinates of the detected object

Returns:

Array of length 2 - top left X and Y

getBottomRightXY

public double[] getBottomRightXY()

Get the bottom right X/Y coordinates of the detected object

Returns:

Array of length 2 - bottom right X and Y

getPredictedClass

public int getPredictedClass()

Get the index of the predicted class (based on maximum predicted probability)

Returns:

Index of the predicted class (0 to nClasses - 1)

toString

public String toString()

Overrides:

toString in class Object