Batch Training

NI Classification Training Interface

Batch Training

Batch Results Batch Classify

Use Tools»Batch Training to train a folder of images.

Complete the following steps to train a folder of images:

  1. Click Tools»Batch Training.
  2. Click the Browse button to select the folder of images to train. After you have selected the folder of images, click Select Cur Dir.
  3. Click Draw ROI to select an ROI around the sample to train.
    Note  Make sure the ROI encompasses the sample to train in all images in the folder.
    Note  If you do not select an ROI in the image, the largest sample will be trained.
  4. Specify labels for the samples to add.
    • If all images in the folder are of the same class, enable the Use the same label for all the samples option, and enter the label in the Class Label control.
    • If the images are not of the same class, enable the Use a file that specifies the labels of the samples option to specify individual labels for each image. Click the Browse button to select a Labels File Path. This file is a delimited text file that contains a corresponding label for each image. A delimiter separates the image name from its label.
  5. Click Train to train the samples in all the images of the folder. The training process trains and classifies the samples.

A dialog box displays the results of the batch training. The first tab displays the label and scores for each sample as well as if it was properly classified. The second tab displays the score histogram, which displays the number of each sample for each score value. The third tab displays the number of samples, standard deviation, accuracy, and predictive value for each class. The fourth tab displays the classification distribution. The last tab displays the mean distance from each class to each other class.