Artificial neural network models are becoming very attractive in image processing where high computational performance and parallel architectures are required. Recently, many papers appeared on applications of neural networks to problems where some degree of intelligence or human-like performance is desired. This paper describes a novel neural network architecture for image recognition and classification. The proposed neural network, called an image recognition neural network (IRNN), is designed to recognize an object or to estimate an attribute of an object. IRNN takes an analog gray level image as an input and produces an appropriate recognition code at the output.