Abstract: Greedy Non-Maximum Suppression (NMS) is a necessity in the object detection pipelines of state-of-the-art models. Because this greedy heuristic can often lead to false negatives when more ...