A white paper on using rendered target objects and real-world environments to build better training datasets for object detection.