Algorithm 1 EIOU-NMS.

Input:  $B=\{b_1,b_2,\dots ,b_n\},S=\{S_1,S_2,\dots ,S_n\},N_t$ B: List with all prediction boxes S: Confidence values corresponding to all prediction boxes $N_t$: Confidence threshold Output:    The final list of prediction boxes D     Implementation process: Step1: Initialize the prediction box output list $D=\Phi$ Step2: Sort S and set the prediction box with the highest confidence as the base box M Step3: Take M from B and put it into D Step4: Iterate through list B and remove the extra prediction boxes using the following $\mathrm{IOU}(M,b_i)-R_{EIOU}(M,b_i)\ge N_t$, then $B=B-b_i,S=S-S_i$  end Step5: Repeat steps 1 through 4 to output the list of prediction boxes D when B is empty