Abstract: Labeling protocols for training datasets and systems and methods for classifying objects of interest and identifying backgrounds. Labeling protocols described herein enable grid units within an image to be excluded from contributing to a loss function, and further enable granular controls for specifying which grid units should contribute to object predictions. A method partitioning an input image into a plurality of grid units and individually processing each of the plurality of grid units with a neural network configured to calculate a confidence score indicating a likelihood that a grid unit comprises an object of interest. The method includes generating a bounding box around the object of interest, identifying one or more grid units of the plurality of grid units that comprise a portion of the bounding box, and identifying which of the one or more grid units comprises a center point of the bounding box.