Convert overlapping bounding boxes to heatmap using predicted probabilities

Question

I am doing object detection based on a sliding window algorithm. Based on the following code I am creating heat maps :

def add_heat(heatmap, bbox_list):