Metallic Powders used in Additive Manufacturing are comprised of particles with varying sizes, morphologies, and features. Most notably of these features are satellites, which are sections of particles in which a smaller particle bonded to it, or a deformation occurred. Until recently, there was not any great way of quantifying or measuring these characteristics. However, recent work by other groups have discovered the usage of a Mask R-CNN for detection and segmentation of satellites from Scanning Electron Microscope (SEM) micrographs is quite effective. Our team has built upon this idea and improved it to be more generalizable, now including 5 different powder types at any magnification between 250x to 1250x. In addition, we formulated a method to expedite the training of a model by implementing model prediction on new images to reduce annotation time. Once each satellite is detected and segmented, its features can be used to identify further relationships between satellites and their conjoined particles. Current methods identify satellites with approximately 80% accuracy for five powder types of magnifications varying from 250x to 1250x. Future work will include fine-tuning this method to improve accuracy as well as relating satellite and particle characteristics and relating them to flowability values of Metallic Powders.