A machine learning algorithm based on a convolutional neural network (CNN) is trained with light microscope images of DLIP patterns to detect the surface homogeneity. 600 images of DLIP patterns with different quantity of homogeneity with line-like structures with spatial periods of 7 µm and different pattern conditions were labelled and used to train the model. The patterns were fabricated using a pulsed laser source with a pulse duration of 4 ns at a wavelength of 1053 nm on stainless steel substrates. For each spatial period, the pulse-to-pulse overlap and laser fluence were varied to obtain more or less homogeneous patterns. Microscopic images were taken of each pattern using a 20x magnification objective. The algorithm is able to detect and identify homogeneous DLIP patterns with an accuracy of 90 %.