Gestalt recognition: beyond basic patterns - descriptors

With the aim of extending the descriptive palette for images, and in particular for patterns, we analysed part of the Gold Fyfe dataset (DOI: 10.5281/zenodo.11032266) following three different strategies. The difference between them is the grade of relatability from a human perspective: the Gestalt Classifier (focused in human kinaesthetic ability to enagage with images), Basic Classifier (retaining aspects of computer and human vision) and traditional image analysis descriptors (computer vision). The classifiers were generated with supervised machine learning, the traditional image analysis descriptors (i.e. Traditional descriptors) were calculated using Fiji ImageJ.

The three data files contains the outcome values for the corresponding image analysis strategy.