Finding the coordinates of an object in an image using dl4j

Question

I started studying neural networks together with the dl4j framework. At the moment I am trying to make the first convolutional network. I have my own dataset where the file name specifies the coordinates of the object and the width with height (x,y,w,h). This is what I want to get as a result. Now I have difficulties in composing the output layer. I don't quite understand what I should get if the object I'm looking for is not found, the second thing I want is to get an integer value of the coordinate or size without scaling it. An example of my neural network is presented below:

  MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder()
            .seed(rngseed)
            .updater(new Adam(0.01))
            .activation(Activation.IDENTITY)
            .weightInit(WeightInit.XAVIER)
            .list()
            .layer(new ConvolutionLayer.Builder(new int[] {5, 5}, new int[] {1, 1}, new int[]{0, 0}).name("cnn1").nIn(1).nOut(64).biasInit(0).build())
            .layer(new SubsamplingLayer.Builder(new int[] {2, 2}, new int[] {2, 2}).name("maxpool1").build())
            .layer(new ConvolutionLayer.Builder(new int[] {5, 5}, new int[] {1, 1}, new int[]{0, 0}).name("cnn2").nIn(64).nOut(16).biasInit(0).build())
            .layer(new OutputLayer.Builder(LossFunctions.LossFunction.NEGATIVELOGLIKELIHOOD)
                    .nOut(4)
                    .activation(Activation.TANH)
                    .build())
            .setInputType(InputType.convolutional(60, 200, 1))
            .build();

Build data set and study:

  for (int e = 0; e < numEpochs; e++) {
        File[] labels = trainData.listFiles();
        for (int i = 0; i < labels.length; i++) {
            File imageFile = labels[i];



            BufferedImage image = ImageIO.read(imageFile);

            INDArray input = loader.asMatrix(image).reshape(1, channels, height, width);
            scaler.fit(new DataSet(input, null));
            scaler.transform(input);

            String name = imageFile.getName();
            int x = Integer.parseInt(name.substring(1,name.indexOf('y')));
            int y = Integer.parseInt(name.substring(name.indexOf('y')+1,name.indexOf('w')));
            int w = Integer.parseInt(name.substring(name.indexOf('w')+1,name.indexOf('h')));
            int h = Integer.parseInt(name.substring(name.indexOf('h')+1,name.lastIndexOf('.')));
            double array[][] ={ { x, y,w,h}};
                    
                    
                
            INDArray output = Nd4j.create(array);

            network.fit(input, output);

        }
    }