Algorithm 1 Multi-Object Selection for Matching.

Require: $N_k,I^Q,{\mathbf{DB}}^L,f$         ▹ # of Objects to pick, Query img, Library & NN

Ensure: ${\mathbf{F}}^{Q^{\prime }},{\mathbf{F}}^{L^{\prime }}$         ▹ Object query & library features selected for matching

1: ${\mathbf{Obj}}^Q,{\mathbf{F}}^Q\leftarrow \mathrm{DYNASCALE}2(f,I^Q)$         ▹ Detect objects & Extract features

2: ${\mathbf{F}}^L,\mathbf{S}\leftarrow \varphi$

3: for  $(Ob{j}^Q,{\mathcal{F}}^Q)\mathbf{in}({\mathbf{Obj}}^Q,{\mathbf{F}}^Q)$  do

4:    ${NF}^Q\leftarrow \#$FAST$\left({\mathcal{F}}_0^Q\right)$         ▹ Get #FAST from RGB of object region

5:    ${\mathcal{F}}^L,{NF}^L\leftarrow \mathrm{GETFEATURES}({\mathbf{DB}}^L,\mathrm{GETCLASSID}\left(Ob{j}^Q\right))$

6:    $\mathbf{S}\leftarrow \mathbf{S}+\left\{|{NF}^Q-{NF}^L|\right\}$

7:    ${\mathbf{F}}^L\leftarrow {\mathbf{F}}^L+\left\{{\mathcal{F}}^L\right\}$

8: end for

9: ${\mathbf{S}}^{\prime }\leftarrow \{s^{\prime }\in \mathbf{S}\mid s^{\prime }\le min(t\in \mathbf{S}\mathrm{such}\mathrm{that}\#\{s\in \mathbf{S}\mid t\le s\}=N_k)\}$

10: ${\mathbf{F}}^{Q^{\prime }},{\mathbf{F}}^{L^{\prime }}\leftarrow \varphi$

11: for  $s\mathbf{in}{\mathbf{S}}^{\prime }$ do

12:    $i\leftarrow \mathrm{index}(s,\mathbf{S})$

13:    ${\mathbf{F}}^{Q^{\prime }}\leftarrow {\mathbf{F}}^{Q^{\prime }}+\left\{{\mathbf{F}}_i^Q\right\}$

14:    ${\mathbf{F}}^{L^{\prime }}\leftarrow {\mathbf{F}}^{L^{\prime }}+\left\{{\mathbf{F}}_i^L\right\}$

15: end for