A Multibranch Object Detection Method for Traffic Scenes

Table 1

The network parameters of MBNet.

Layer	Type	Filters	Size/stride	Input	Output
0	Conv	16	3 ∗ 3/1	512 ∗ 512 ∗ 3	512 ∗ 512 ∗ 16
1	Maxpool		2 ∗ 2/2	512 ∗ 512 ∗ 16	256 ∗ 256 ∗ 16
2	Conv	32	3 ∗ 3/1	256 ∗ 256 ∗ 16	256 ∗ 256 ∗ 32
3	Maxpool		2 ∗ 2/2	256 ∗ 256 ∗ 32	128 ∗ 128 ∗ 32
4	Conv	64	3 ∗ 3/1	128 ∗ 128 ∗ 32	128 ∗ 128 ∗ 64
5	Maxpool		2 ∗ 2/2	128 ∗ 128 ∗ 64	64 ∗ 64 ∗ 64
6	Conv	128	3 ∗ 3/1	64 ∗ 64 ∗ 64	64 ∗ 64 ∗ 128
7	Maxpool		2 ∗ 2/2	64 ∗ 64 ∗ 128	32 ∗ 32 ∗ 128
8	Conv	256	3 ∗ 3/1	32 ∗ 32 ∗ 128	32 ∗ 32 ∗ 256
9	Maxpool		2 ∗ 2/2	32 ∗ 32 ∗ 256	16 ∗ 16 ∗ 256
10	Conv	512	3 ∗ 3/1	16 ∗ 16 ∗ 256	16 ∗ 16 ∗ 512
11	Maxpool		2 ∗ 2/1	16 ∗ 16 ∗ 512	16 ∗ 16 ∗ 512
12	Conv	1024	3 ∗ 3/1	16 ∗ 16 ∗ 512	16 ∗ 16 ∗ 1024
13	Conv	256	1 ∗ 1/1	16 ∗ 16 ∗ 1024	16 ∗ 16 ∗ 256
14	Conv	512	3 ∗ 3/1	16 ∗ 16 ∗ 256	16 ∗ 16 ∗ 512
15	Conv	128	3 ∗ 3/1	16 ∗ 16 ∗ 512	16 ∗ 16 ∗ 128
16	Conv	36	1 ∗ 1/1	16 ∗ 16 ∗ 128	16 ∗ 16 ∗ 36
17	Yolo
18	Route 14
19	Conv	128	1 ∗ 1/1	16 ∗ 16 ∗ 512	16 ∗ 16 ∗ 128
20	Upsample		2×	16 ∗ 16 ∗ 128	32 ∗ 32 ∗ 128
21	Route 20 8
22	Conv	256	3 ∗ 3/1	32 ∗ 32 ∗ 384	32 ∗ 32 ∗ 256
23	Conv	36	1 ∗ 1/1	32 ∗ 32 ∗ 256	32 ∗ 32 ∗ 36
24	Yolo
25	Route 22
26	Conv	128	1 ∗ 1/1	32 ∗ 32 ∗ 256	32 ∗ 32 ∗ 128
27	Upsample		2×	32 ∗ 32 ∗ 128	64 ∗ 64 ∗ 128
28	Route 27 6
29	Conv	512	1 ∗ 1/1	64 ∗ 64 ∗ 256	64 ∗ 64 ∗ 512
30	Conv	36	1 ∗ 1/1	64 ∗ 64 ∗ 512	64 ∗ 64 ∗ 36
31	Yolo