Process monitoring in additive manufacturing should allow components to be certified cheaply and rapidly and opens the possibility of healing defects, if detected. Here, neural networks (NNs) and convolutional neural networks (CNNs) are trained to detect flaws in layerwise images of a build, using labeled XCT data as a ground truth. Multiple images were recorded after each layer before and after recoat with various lighting conditions. Classifying networks were given a single image or multiple images of various lighting conditions for training and testing. CNNs demonstrated significantly better performance than NNs when testing and training on data from the same component. CNNs also performed better when training on all data available from one build, and testing on data from an unseen build. CNNs demonstrated accuracies of 93% when testing and training within the same component, and 79% when testing on a previously unseen build. CNNs were demonstrated to have superior generalizability compared to NNs. As well, data fusion techniques were shown to raise the out of class accuracy to 83%. It was determined that the size of voids was a strong determining factor in whether they can be detected with these classifiers in layerwise imagery; classifiers trained on only large voids were able to achieve out of class accuracy of 87%.