Progressive multi-granularity ResNet vehicle recognition network

Fig. 5. Top1/% curve of change. (a) Effect of β values on RCDB on Stanford-cars; (b) Effect of β values on RCDB on Compcars; (c) Effect of β values on RCDB on VMRURS

Download full size

View in Article

Fig. 6. Network training and testing process

Download full size

View in Article

Fig. 7. Visual comparison of vehicle recognition in each stage

Download full size

View in Article

Fig. 8. Visual comparison of after adding each module

Download full size

View in Article

Fig. 9. Visual comparison of different network vehicle recognition

Download full size

View in Article

Table 1. Progressive multi-granularity training steps

View table

View in Article

Table 1. Progressive multi-granularity training steps

渐进式多粒度训练过程
输入：训练数据集D，训练数据的批次为x，标签样本为y，P代表多粒度渐进式网络学习， $L_{CE}$ 代表交叉熵损失(cross entropy loss, CE)
$For e p o c h \in [0, e p o c h s] do$
$For b \in [0, b a t c h s] do$
$x, y \Leftarrow b a t c h b of D$
$For l \in [L - S + 1, L] do$
$y^{l} \Leftarrow H_{class}^{l} [H_{Conv}^{l} (F^{l} (P (x, n)))]$
$L_{l} \Leftarrow L_{CE} (y^{l}, y)$
Backpropagation $L_{l}$
End for
$y^{Concat} = H_{class}^{Concat} {Concat [V^{(L - S + 1)}, \dots, V^{L}]}$
$L_{Concat} \Leftarrow L_{CE} (y^{Concat}, y)$
Backpropagation $L_{Concat}$
End for
End for

Table 2. Comparison of PLCB split size at each stage
View table
View in Article
Table 2. Comparison of PLCB split size at each stage
Stage1 Stage2 Stage3 Stage4 Accuracy/%
1 1 1 1 93.5
2 2 2 2 93.9
4 4 4 4 94.2
8 8 8 8 94.0
16 8 4 2 93.6
8 4 2 1 94.5
4 2 1 1 93.9

Table 3. Ablation experiment of recognition effect after adding different layers to the RCDB module
View table
View in Article
Table 3. Ablation experiment of recognition effect after adding different layers to the RCDB module
ResNet50 Layer1 Layer2 Layer3 Layer4 Accuracy/%
✓ ✗ ✗ ✗ ✗ 91.5
✓ ✓ ✗ ✗ ✗ 92.3
✓ ✗ ✓ ✗ ✗ 92.9
✓ ✗ ✗ ✓ ✗ 93.0
✓ ✗ ✗ ✗ ✓ 92.6
✓ ✓ ✓ ✓ ✓ 93.2

Table 4. Different modules are added to the network
View table
View in Article
Table 4. Different modules are added to the network
Baseline PLCB RCDB Accuracy/%
✓ ✗ ✗ 91.5
✓ ✓ ✗ 94.8
✓ ✗ ✓ 93.2
✓ ✓ ✓ 95.7

Table 5. Comparison of recognition accuracy of different network models

View table

View in Article

Table 5. Comparison of recognition accuracy of different network models

Methods	Backbone	Stanford-cars/%	Compcars/%	VMRURS/%	Speed/(f/s)	Params/M	FLOPs/G
基线^[25]	ResNet50	91.5	94.1	87.1	4.15	23.50	33.05
FBSD^[31]	ResNet50	94.4	96.8	92.3	1.73	46.82	53.11
LIO^[32]	ResNet50	94.5	96.8	94.2	3.60	24.57	33.06
DCL^[33]	ResNet50	94.5	96.7	94.7	3.46	24.91	33.06
Cross-X^[34]	ResNet50	94.6	97.0	94.6	3.88	25.56	38.86
CAL^[17]	ResNet50	95.5	98.0	96.4	3.72	33.73	33.08
WS-DAN^[18]	ResNet50	94.5	97.1	95.6	4.02	33.24	33.08
PMG^[26]	ResNet50	95.1	97.8	95.7	2.94	45.12	69.82
CN-CNN^[35]	ResNet50	94.9	97.6	94.9	1.92	42.31	47.65
Ours	ResNet50	95.7	98.8	97.4	2.97	40.64	69.61

Tools

Get Citation

Copy Citation Text

Shengjun Xu, Yang Jing, Haitao Li, Zhongxing Duan, Fuyou Liu, Minghai Li. Progressive multi-granularity ResNet vehicle recognition network[J]. Opto-Electronic Engineering, 2023, 50(7): 230052

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites