Successfully optimized deep learning models to detect 15 distinct objects through implementation of image tiling and innovative Strategic Aerial Homogenization for Inference (SAHI) approach to improve mean average precision by 36%