Identifying two anatomically similar species of Cupressaceae, Chamaecyparis obtusa and Thujopsis spp., is important to better understand the culture of wood use in Japan. However, the conventional method, which involves observing their cross-field pitting, cannot identify them in many cases. This study solves the above problem by introducing an anatomical criterion based on the micro fibril angle (MFA). MFA values were obtained through two-dimensional MFA images using the uniaxial optical anisotropy of cellulose microfibrils. A combination of the preprocessed MFA images and a convolutional neural network (CNN) yielded an accuracy nearly of 90% in classifying these species in cases of present and old wood specimens. Our feature extraction and classification techniques provide a new way for describing the anatomical features of wood and identifying featureless softwoods. Using the model interpretation-related methodologies of the CNN, distinct features of the two wood species were partly explained by MFA anisotropy in the S2 wall induced by the existence of pits.