With enhances in materials, especially carbon and glass fibers, the use of short fiber composites increased in the last decade. Moreover, functionality of these materials boosts because of these advances. Nowadays, short fiber composites show better performance and are used in more structural and functional applications. A great percentage of these type of composites are manufactured with the injection molding method. This method is superior to other methods of manufacturing composites because of its privileges in forming most kind of shapes and profiles. However, the process of manufacturing these composites results in arbitrary positioning of fibers in matrix due to fluid flow during manufacturing. Composite materials became an outstanding substitution for traditional materials, as they present a better mechanical performance. Injection molded composites gained a great enthusiasm in different applications, specifically in automotive and aerospace industries as they are practically formable to any profile. Because of their manufacturing method, they form a random orientation of fibers in resin. Orientation of fibers is the most dominant parameter influencing the mechanical properties of composites and needs to be studied to gain a safe prediction for properties of manufactured properties. As a result, it is essential to predict or measure fiber orientation accurately for this purpose. However, in most of the studies random geometry of composites is approximated and simplified. In this study, an objected oriented finite element method has been presented which models composites by the finite element method based on their real orientations. Randomly oriented composite sheets of glass and epoxy have been produced, and based on their scanned images the finite element mesh has been generated. Finally a prediction of Elastic modulus of these specimens has been made and compared with experimental data.