Zhuoran Jiang

Ms. Zhuoran Jiang

Nanjing Normal University, Nanjing, Jiangsu, China
orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
Verified
Journal Article
Open access
PDF file
ZJZhuoran Jiang
Zhuoran Jiang
XZXinxin Zhou
Xinxin Zhou
WCWei Cao
Wei Cao
ZSZaihong Sun
Zaihong Sun
CWChangbin Wu
Changbin Wu
In recent years, deep learning has become the mainstream development direction in the change-detection field, and its accuracy and speed have also reached a high level. However, the change-detection method based on deep learning cannot predict all the change areas accurately, and its application is limited due to local prediction defects. For this reason, we propose an interactive change-detection network (ICD) for very high resolution (VHR) based on a deep convolution neural network. The network integrates positive- and negative-click information in the distance layer of the change-detection network, and users can correct the prediction defects by adding clicks. We carried out experiments on the open source dataset WHU and LEVIR-CD. By adding clicks, their F1-scores can reach 0.920 and 0.912, respectively, which are 4.3% and 4.2% higher than the original network. To better evaluate the correction ability of clicks, we propose a set of evaluation indices—click-correction ranges, which is suitable for evaluating clicks, and we carry out experiments on the above models. The results show that the method of adding clicks can effectively correct the prediction defects and improve the result accuracy.

22 February 2023 in Land
Verified
Journal Article
Open access
PDF file
ZJZhuoran Jiang
Zhuoran Jiang
MJMing Jiang
Ming Jiang
YWYahua Wang
Yahua Wang
CMCan Ma
Can Ma
WQWeifeng Qiao
Weifeng Qiao
As the largest developing country, China has permanently attached great importance to cultivated land protection. However, due to the different rules of cultivated land identification in the second and third national land surveys, the cultivated land area in the two surveys has changed greatly. Some agricultural lands in the south, such as plantations, forests, grasslands, aquaculture ponds, etc., belonged to cultivated land during the second survey, but they were identified as non-cultivated land in the third national land survey. This change has led to a sharp reduction in the area of cultivated land in some places. In order to calculate the actual change in the area of cultivated land since the second survey and provide a reasonable basis for the standard of cultivated land protection, this paper takes Gaochun District, a developed area in China, as an example; interprets the images of the second national land survey period with the deep learning network HRNet; and compares the results with the second and third national land survey rules. The results show that the actual reduction of cultivated land in Gaochun District in the past ten years accounts for 35.1% of the reduction of cultivated land in the two land surveys, while the reduction of cultivated land caused by the change of cultivated land identification rules accounts for 64.9% of the reduction of cultivated land in the two land surveys, indicating that the significant reduction in local cultivated land was mainly caused by the changes in the rules, and these cultivated land reduction behaviors existed before the second survey.

25 April 2023 in TechRxiv
Verified
Preprint
Open access
PDF file
ZJZhuoran Jiang
Zhuoran Jiang
XZXinxin Zhou
Xinxin Zhou
WCWei Cao
Wei Cao
ZSZaihong Sun
Zaihong Sun
CWChangbin Wu
Changbin Wu
In recent years, deep learning has become the mainstream development direction in the change detection field, and its accuracy and speed have also reached a high level. However, the change detection method based on deep learning cannot predict all the change areas accurately, and its application is limited due to local prediction defects. For this reason, we propose an interactive change detection network (ICD) for very high resolution (VHR) based on a deep convolution neural network. The network integrates positive and negative click information in the distance layer of the change detection network, and users can correct the prediction defects by adding clicks. We carried out experiments on the open source dataset WHU and LEVIR-CD. By adding clicks, their F1-scores can reach 0.920 and 0.912, respectively, which are 4.3% and 4.2% higher than the original network. To better evaluate the correction ability of clicks, we propose a set of evaluation index—click correction ranges, which is suitable for evaluating clicks, and we carry out experiments on the above models. The results show that the method of adding clicks can effectively correct the prediction defects and improve the result accuracy.