Algorithm 2: The pseudocode of non-maximum suppression (NMS) for our approach

Original Bounding Boxes:   $B=[b_1,\dots ,b_M]$, $C=[c_1,\dots ,c_M]$, $threshold=0.6$   $B$ refers to the set of original bounding boxes   $C$ refers to the set of confidences of $B$ Detection result:   $F$ refers to the set of the final bounding boxes 1:  $F\leftarrow []$ 2:   while $B\ne []$ do: 3:     $k\leftarrow \arg \max C$ 4:     $F\leftarrow F.append(b_k)$; $B\leftarrow delB[b_k]$; $C\leftarrow delC[c_k]$ 5:     for $b_i\in B$ do: 6:       if $\mathrm{IOU}(b_i,b_k)\ge threshold$ 7:        $B\leftarrow delB[b_i]$; $C\leftarrow delC[c_i]$ 8:       end 9:     end 10:  end