MDESNet: Multitask Difference-Enhanced Siamese Network for Building Change Detection in High-Resolution Remote Sensing Images

Round 1

Reviewer 1 Report

Much improved version from the past submission.

Reviewer 2 Report

The authors have answered all of my concerns.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Good experiments. However I have the following concerns:

1. Table 5 shown UNET++ provides better results, then why use the new approach?

2. The study requires detailed ablation study results showing how each stage improves claimed performance.

3. Highlight which part is new in the architecture.

4. The wring is not clear throughout the manuscript with complex and sometimes obscure sentences where meaning of the sentence is lost. I would highly recommend through English editing by an expert.

 

Reviewer 2 Report

For the task of change detection in high-resolution remote sensing images, this manuscript proposed a multi-task difference enhance siamese network by introducing semantic constraints and feature difference enhance module. The idea is interesting and the result is basically satisfactory. However, some other problems in the manuscript are still concerned in the following:

1. The proposed method was validated on the building change detection dataset. In fact, building change detection is an easy branch of change detection. More kinds of public datasets should be tested in the experiments.

2. As shown in Table 4, the proposed method didn’t have advantage on the recall, why?

3. In Figure 9, could the authors show more comparison methods?

4. Could the authors add a weight for a trade-off between binary cross-entropy loss function and the focal loss function in Eq. (5)?

5. More recent works on change detection could be included, such as “DOI: 10.1109/IGARSS47720.2021.9554522”, “DOI: 10.1109/LGRS.2021.3098774”...

6. I suggest the authors to add a keyword of “remote sensing” or “high-resolution remote sensing”