Building Change Detection for Aerial Images Based on Attention Pyramid Network

Qinglin Tian; Kai Qin; Jun Chen; Yao Li; Xuejiao Chen

doi:10.3788/AOS202040.2110002

Journals >Acta Optica Sinica >Volume 40 >Issue 21 >Page 2110002 > Article

Acta Optica Sinica
Vol. 40, Issue 21, 2110002 (2020)

Building Change Detection for Aerial Images Based on Attention Pyramid Network

Qinglin Tian^1、*, Kai Qin¹, Jun Chen², Yao Li³, and Xuejiao Chen¹

Author Affiliations

¹National Key Laboratory of Remote Sensing Information and Image Analysis Technology, Beijing Research Institute of Uranium Geology, Beijing 100029, China

²Iflytek Intelligent Information Technology Co., Ltd., Hefei, Anhui 230094, China

³Zachry Department of Civil and Environmental Engineering, Texas A & M University, Texas 77843, USA;

show less

DOI: 10.3788/AOS202040.2110002 Cite this Article Set citation alerts

Qinglin Tian, Kai Qin, Jun Chen, Yao Li, Xuejiao Chen. Building Change Detection for Aerial Images Based on Attention Pyramid Network[J]. Acta Optica Sinica, 2020, 40(21): 2110002 Copy Citation Text

show less

Fig. 1. Diagrams of (a) traditional convolution and (b) dilated convolution. (a) Traditional convolution; (b) dilated convolution

Download full size

Fig. 2. Structural diagram of PPM

Download full size

Fig. 3. Structural diagram of CBAM

Download full size

Fig. 4. Illustration of network architecture

Download full size

Fig. 5. Fusion of features with different scales

Download full size

Fig. 6. Illustration of data augmentation

Download full size

Fig. 7. Comparison of results before and after post-processing by proposed method

Download full size

Fig. 8. Comparison of examples of network ablation models. (a) Image 1; (b) image 2; (c) ground truth; (d) proposed network; (e) ablation model A; (f) ablation model B; (g) ablation model C; (h) enlarged drawing in box

Download full size

Fig. 9. Results of detection for ordered building change. (a) Image 1; (b) image 2; (c) ground truth; (d) UNet; (e) DeepLab; (f) CSCDNet; (g) UPerNet; (h) proposed network

Download full size

Fig. 10. Results of detection for multi-scale building change. (a) Ground truth; (b) UNet;(c) DeepLab; (d) CSCDNet; (e) UPerNet; (f) proposed network

Download full size

ResNet101 convolutional layer	Stage name	Output feature	Output scale
7×7,64,stride 2	conv1	C₁	1/2
3×3,max pooling,stride 2
$(\begin{array}{l} 1 \times 1,64 \\ 3 \times 3,64 \\ 1 \times 1,256 \end{array})$ ×3	conv2	C₂	1/4
$(\begin{array}{l} 1 \times 1,128 \\ 3 \times 3,128 \\ 1 \times 1,512 \end{array})$ ×4	conv3	C₃	1/8
$(\begin{array}{l} 1 \times 1,256 \\ 3 \times 3,256 \\ 1 \times 1,1024 \end{array})$ ×23(DC)	conv4	C₄	1/8
$(\begin{array}{l} 1 \times 1,512 \\ 3 \times 3,512 \\ 1 \times 1,2048 \end{array})$ ×3 (DC)	conv5	C₅	1/8

Table 1. Architecture of feature extraction network in encoding stage

Ablation experiment	DC	CBAM	PPM	P /%	R /%	F₁ /%
A	O	P	P	85.16	83.10	84.11
B	P	O	P	83.94	87.07	85.47
C	P	P	O	84.54	86.71	85.61
Ours	P	P	P	84.47	88.10	86.25

Table 2. Ablation experiment analysis of network

Method	P /%	R /%	F₁ /%
UNet	73.09	42.84	54.02
DeepLab	77.73	51.41	61.89
CSCDNet	81.08	69.60	74.90
UPerNet	78.66	71.99	75.18
Ours	84.47	88.10	86.25

Table 3. Accuracy assessment of building change detection by different methods

Qinglin Tian, Kai Qin, Jun Chen, Yao Li, Xuejiao Chen. Building Change Detection for Aerial Images Based on Attention Pyramid Network[J]. Acta Optica Sinica, 2020, 40(21): 2110002

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information