Two-Stream Swin Transformer with Differentiable Sobel Operator for Remote Sensing Image Classification

Round 1

Reviewer 1 Report

This paper has presented a two-stream swin transformer network (TSTNet) for remote sensing scene classification. In the network, the original and edge features extracted from the original and edge streams are fused to make classification. Besides, a differentiable edge Sobel operator module is included in the edge stream which can learn the parameters of Sobel operator adaptively. Generally, the paper is well organized, and the proposed method is verified to be promising.

1-One of the major contributions is adopting Sobel operator for edge information extracting. The related works on using edge curves for remote sensing scene classification should be further reviewed.

2-A learnable Sobel operator is developed and used in TSTNet. Why choose Sobel operator, and how about the other types of edge extraction operators, such as the Canny operator? One of my concerns is the rationality of the using of Sobel operator.

3-The feature for remote sensing scene classification should be invariant of transformation, rotation, and scale. Whether the proposed TSTNet has such properties for remote sensing scene feature learning and classification?

4-It’s recommended to include the discussion on the overall computational complexity or running time for the proposed TSTNet.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Overall, it's a well-organized study and its results.

The expression remote sensing scene classification is a somewhat awkward expression. Remote sensing image classification is more common, so I recommend modification.

Accuracy is emphasized, but actual speed is also an important factor. Performance comparison for speed is required.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have addressed all the problems and the manuscript can be accepted for publication. 

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.