Algorithm 1. Set Up Maximum and Minimum Bounding Boxes for Each Section

Input:    - ‘bg_imgs’: List of background images    - ‘bg_txts’: List of text files corresponding to ‘bg_imgs’ Output:    - ‘results’: Dictionary of containing bounding box statistics

For each ‘bg_img’ in ‘bg_imgs’ do    Open the corresponding file in ‘bg_txts’ and read lines into ‘bg_lines’    For each ‘bg_line’ in ‘bg_lines’ do      Parse ‘x’, ‘y’, ‘w’, ‘h’ as floats from ‘bg_line’      Adjust ‘y’ to (‘y’ − ‘h’/2)      Create ‘bbox’ as a tensor [‘x’, ‘y’, ‘w’, ‘h’]      Calculate ‘area’ as ‘w’ * ‘h’      Increment ‘count’      Update bounding box statistics:        ‘ymin’, ‘ymax’, ‘bboxmin’, ‘bboxmax’ based on ‘y’        ‘areamax’, ‘widthmax’, ‘heightmax’, ‘widthmin’, ‘heightmin’ based on ‘w’, ‘h’, and ‘area’      If ‘prev_bbox’ exists, then        Calculate ‘iou’        If ‘iou’ equals 0, then          Store current statistics in ‘results’ for ‘section_num’          Reset statistics          Increment ‘section_num’      Set ‘prev_bbox’ to ‘bbox’    Store final statistics in ‘results’ for the last ‘section_num’ Return ‘results’