High-Resolution Remote Sensing Image Classification Based on DeeplabV3+ Network

Fig. 7. Comparison of the classification results of the four models. (a) Original image; (b) label; (c) MDeeplabV3+; (d) XDeeplabV3+;(e) MS-MDeeplabV3+; (f) MS-XDeeplabV3+

Download full size

Table 1. Similarities and differences between the four network structures

View table

Table 1. Similarities and differences between the four network structures

Model structure	Encoder		Decoder
Model structure	MobilenetV2	Xception_65	Skip a layer fusion	Layer-by-layer fusion	Channel module	Multi-scale supervision
MDeeplabV3+	√		√
XDeeplabV3+		√	√
MS-MDeeplabV3+	√			√	√	√
MS-XDeeplabV3+		√		√	√	√

Table 2. Detailed configuration of the MobilenetV2
View table
Table 2. Detailed configuration of the MobilenetV2
Input size Operation t c n s
256²×3 Conv2d 32 1 2
128²×32 Bottleneck 1 16 1 1
128²×16 Bottleneck 6 24 2 2
64²×24 Bottleneck 6 32 3 2
32²×32 Bottleneck 6 64 4 2
16²×64 Bottleneck 6 96 3 1
16²×96 Bottleneck 6 160 3 1
16²×160 Bottleneck 6 320 1 1

Table 3. Proportion of each category in the CCF dataset
View table
Table 3. Proportion of each category in the CCF dataset
Parameter Building Arable Forest Water Road Grass Other
Label 0 1 2 3 4 5 6
Percentage /% 2.79 50.87 17.87 17.74 0.35 1.96 7.38

Table 4. Quantitative evaluation of the classification results of each model

View table

Table 4. Quantitative evaluation of the classification results of each model

Method	IoU							mIoU	OA	Kappa
Method	Building	Arable	Forest	Water	Road	Grass	Other	mIoU	OA	Kappa
MDeeplabV3+	0.6882	0.8047	0.7828	0.8249	0.2015	0.2286	0.6233	0.5934	0.8027	0.7735
XDeeplabV3+	0.7144	0.8240	0.8267	0.8706	0.2554	0.2034	0.6480	0.6204	0.8348	0.7992
MS-MDeeplabV3+	0.7217	0.8234	0.8186	0.8625	0.3431	0.2298	0.6442	0.6348	0.8502	0.8297
MS-XDeeplabV3+	0.7650	0.8836	0.8437	0.9168	0.4774	0.3309	0.6615	0.6970	0.9122	0.8646

Table 5. Comparison of different network models' training results

View table

Table 5. Comparison of different network models' training results

Methed	Parameter size /MB	Time /min	mIoU
FCN	364.7	37.8	0.5586
U-Net	305.3	33.6	0.5739
SegNet	307.6	34.5	0.5748
DeeplabV3+	312.2	39.1	0.6199
E-Deeplab	387.3	47.8	0.6835
Algorithm in Ref.［18］	246.4	37.2	0.6621
Algorithm in Ref.［19］	332.6	42.4	0.6537
MDeeplabV3+	52.7	13.3	0.5934
XDeeplabV3+	148.5	18.7	0.6204
MS-MDeeplabV3+	55.3	17.6	0.6348
MS-XDeeplabV3+	151.1	23.9	0.6970

Table 6. Comparison of different loss weights for
View table
Table 6. Comparison of different loss weights for
ID Weight of side output mIoU OA Kappa
D₁ D₂ D₃ D₄
1 0 0 0 1 0.6159 0.8214 0.7975
2 0.3 0.3 0.8 1 0.6317 0.8485 0.8283
3 1 1 1 1 0.6348 0.8502 0.8297

Table 7. Comparison of different loss weights for
View table
Table 7. Comparison of different loss weights for
ID Weight of side output mIoU OA Kappa
D₁ D₂ D₃ D₄
1 0 0 0 1 0.6681 0.8827 0.8447
2 0.3 0.3 0.8 1 0.6933 0.9096 0.8629
3 1 1 1 1 0.6970 0.9122 0.8646

Tools

Get Citation

Copy Citation Text

Dongqing Huang, Weiming Xu, Wendi Xu, Xiaoying He, Kaixiang Pan. High-Resolution Remote Sensing Image Classification Based on DeeplabV3+ Network[J]. Laser & Optoelectronics Progress, 2023, 60(16): 1628001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites