A variation of the neural algorithm locally excitatory globally inhibitory network has been developed for automatic recognition, count, and morphological analysis of a large number of quantum dots. In a typical image of 256 x 256 pixels (500 x 500 nm(2)), the program recognizes the dots with an average error of about 3%. It also calculates area and height of each of them. Furthermore, the recognition procedure is very fast (less than 2-3 s on a Pentium III at 600 MHz). (C) 2004 American Vacuum Society.