You are to design a visual system for detecting small black objects against a tight background. Assume that the image of one of these objects is a square that is 5 pixels wide. Your system is to be used to create a Boolean feature that has value I whenever a single such image appears anywhere in a 100 x 100 image any and has value 0 for any other image. Describe how you would design such a system, first using a neural network (with one hidden layer) and second, using convolution with a specially designed weighting function.