Segmentation of Aerial Image with
Multi-scale Feature and Attention Model
Shiyu Hu
1 , Qian Ning
1,2(B) , Bingcai Chen
3 , Yinjie Lei
1 , Xinzhi Zhou
1 ,
Hua Yan
1 , Chengping Zhao
1 , Tiantian Tang
2 , and Ruiheng Hu
2
1 Sichuan University, Chengdu, Sichuan, China
hu shiyu@163.com, ningq@scu.edu.cn, yinjie@scu.edu.cn, xz.zhou@scu.edu.cn,
yanhua@scu.edu.cn, sc zcp@scu.edu.cn
2 Xinjiang Normal University, Urumqi, Xinjiang, China
3 Dalian University of Technology, Dalian, China
china@dlut.edu.cn
Abstract. Aerial image labeling plays an important part in the mapping of maps with high precision. The knowledge about the range and
intensive degree of aerial building segmentation is necessary for urban
planning. Fully convolutional networks (FCNs) have recently shown
state-of-the-art performance in image segmentation. In order to get better aerial images segmentation performance, we use a method of combing
FCNs with multi-scale features and attention model in order to carry
out segmentation automatically in aerial images. Attention model gives
each scale feature added extra supervision to achieve better segmentation. Here, U-net and FCN-8s are used as original semantic segmentation model to train with multi-scale images and attention models. The
datasets use different proportions of Inria Aerial Image Labeling Dataset,
including two semantic classes: building and not building. The results
show that the semantic segmentation model combined with multi-scale
features and attention model has higher segmentation accuracy and better performance.
Keywords: Image segmentation · Aerial image labeling · Fully
convolution neural networks · Attention model · Multi-scale feature
1 Introduction
Aerial image labeling contributes to the mapping of land cover and change detection and is used in areas such as forestry and urban planning. Due to the complexity of aerial image data, aerial image labeling has the following challenges:
– Occlusion—Partial or full occlusion can be caused by other objects such as
tree, while partial occlusion which occurs more often is mostly caused by
trees.
c
Springer Nature Singapore Pte Ltd. 2020
Q. Liang et al. (Eds.): Artificial Intelligence in China, LNEE 572, pp. 58–66, 2020.
https://doi.org/10.1007/978-981-15-0187-6_7
Multi-scale Feature and Attention Model
Shiyu Hu
1 , Qian Ning
1,2(B) , Bingcai Chen
3 , Yinjie Lei
1 , Xinzhi Zhou
1 ,
Hua Yan
1 , Chengping Zhao
1 , Tiantian Tang
2 , and Ruiheng Hu
2
1 Sichuan University, Chengdu, Sichuan, China
hu shiyu@163.com, ningq@scu.edu.cn, yinjie@scu.edu.cn, xz.zhou@scu.edu.cn,
yanhua@scu.edu.cn, sc zcp@scu.edu.cn
2 Xinjiang Normal University, Urumqi, Xinjiang, China
3 Dalian University of Technology, Dalian, China
china@dlut.edu.cn
Abstract. Aerial image labeling plays an important part in the mapping of maps with high precision. The knowledge about the range and
intensive degree of aerial building segmentation is necessary for urban
planning. Fully convolutional networks (FCNs) have recently shown
state-of-the-art performance in image segmentation. In order to get better aerial images segmentation performance, we use a method of combing
FCNs with multi-scale features and attention model in order to carry
out segmentation automatically in aerial images. Attention model gives
each scale feature added extra supervision to achieve better segmentation. Here, U-net and FCN-8s are used as original semantic segmentation model to train with multi-scale images and attention models. The
datasets use different proportions of Inria Aerial Image Labeling Dataset,
including two semantic classes: building and not building. The results
show that the semantic segmentation model combined with multi-scale
features and attention model has higher segmentation accuracy and better performance.
Keywords: Image segmentation · Aerial image labeling · Fully
convolution neural networks · Attention model · Multi-scale feature
1 Introduction
Aerial image labeling contributes to the mapping of land cover and change detection and is used in areas such as forestry and urban planning. Due to the complexity of aerial image data, aerial image labeling has the following challenges:
– Occlusion—Partial or full occlusion can be caused by other objects such as
tree, while partial occlusion which occurs more often is mostly caused by
trees.
c
Springer Nature Singapore Pte Ltd. 2020
Q. Liang et al. (Eds.): Artificial Intelligence in China, LNEE 572, pp. 58–66, 2020.
https://doi.org/10.1007/978-981-15-0187-6_7