A Light Vehicle License-Plate-Recognition System Based on Hybrid Edge–Cloud Computing

. 2023 Nov 2;23(21):8913. doi: 10.3390/s23218913

Algorithm 1 Soft-NMS method.

Input:

B = {b_{1}, \dots, b_{N}}, C = {c_{1}, \dots, c_{N}}, T_{n m s}

B is a list of the initial detection boxes
C contains a list of corresponding detection scores

T_{n m s}

relates to is the NMS threshold
Output: Final detection Bounding box list S

1:
function $S O F T_N M S$
2:
while $B \neq e m p t y$ do
3:
$m \leftarrow arg max C$
4:
$S \leftarrow S \cup b_{m}; B \leftarrow B \cup b_{m}; C \leftarrow C \cup c_{m}$
5:
$s_{i} \leftarrow s_{i} f (I o U (M, b_{i}))$
6:
for $b_{i} \in B$ do
7:
if $I o U (M, b_{i}) \geq T_{n m s}$ then
8:
$B \leftarrow B \cup b_{m}; C \leftarrow C \cup c_{m}$
9:
else
10:
$s_{i} \leftarrow s_{i} f (I o U (M, b_{i}))$
11:
end if
12:
end for
13:
end while
14:
return result