This article is Part I of a set of five technical articles that accompany a whitepaper written in collaboration between Meta and Ekimetrics. Object Detection (OD) and Optical Character Recognition (OCR) were used to detect specific features in creative images, such as faces, smiles, text, brand logos, etc. Then, in combination with impressions data, marketing mix models were used to investigate what objects, or combinations of objects in creative images in marketing campaigns, drive higher ROIs. In this Part I we explore the methodology for using pre-trained Detectron2 models to detect brand-specific object in creative images.