Image classification

Image Classifier scans through all folders and sub folders from the base path specified in the Autoclustering options dialog and automatically groups similar documents into classes. Each time a new document is encountered a new class is added and trained with the new document. Each time a document matches with an existing class it is counted to that class and optionally trained to improve quality.

The following buttons are available on the Image Classifier window:

Start

Begins scanning directories and Autoclustering. First all file names in the folders are loaded into the memory, then the clustering starts. Depending on the number of files present in the directory structure loading the file names can take some time. However, it is no problem to cluster 10.000 images and more.

Stop

Click to end scanning and Autoclustering.

Options

Displays the Autoclustering Options window to setup parameters for autoclustering.

The following Statistics information is available when documents are processed:

Document count

The number of images that are already clustered with respect to all images in the folders.

Class count

The number of classes that already are created.

Classification time

The time (in ms) needed to classify the current document.

A progress bar shows the current status of the clustering process.