MFACNet: A Multi-Frame Feature Aggregating and Inter-Feature Correlation Framework for Multi-Object Tracking in Satellite Videos

Round 1

Reviewer 1 Report

The paper is well put together.  Contributions, metrics, and experiment are clearly articulated.

Author Response

Dear Reviewer,

Thank you for your valuable feedback and constructive comments on our manuscript titled [MFACNet: A Multi-frame Features Aggregating and In-ter-Feature Correlation Framework for Multi-Object Tracking in Satellite Videos || Manuscript ID remotesensing-2917758]. We appreciate the time and effort you have put into reviewing our work. We have carefully considered your comments and suggestions, and we have made the necessary revisions to address them.

We believe that the revisions we have made have significantly improved the quality and clarity of our manuscript. We have addressed all the concerns raised by the reviewers and have provided additional explanations and evidence where necessary. We hope that these revisions meet your expectations and address any remaining concerns.

Once again, we would like to express our gratitude for your valuable feedback and for helping us improve our manuscript. We believe that your comments have greatly contributed to the overall quality of our work. We look forward to hearing your feedback on the revised version of our manuscript.

Thank you once again for your time and consideration.

Sincerely,

Hu Zhao

Reviewer 2 Report

The work is well presented and provides a sufficient degree of novelty to be published. Minor revisions required for more efficient presentation.

Line 72: ViBE acronym seems undefined.

Figure 3: I advise making the picture full page and explain better graphically the inputs to the FAW and FMME modules. I understood that they take the very same features input but is not really clear from the figure. Also, I would use the same notation of the manuscript where I understood f^(t-tau) is instead feats(t-tau).

Line 227: the sentence “where w(t−τ) is enable its participation in the training of FAW module” seems grammatically incorrect, please doublecheck.

Line 361: I suggest recalling the precise definition of “Mostly tracked trajectories” and “Mostly lost trajectories” indexes to better interpret the results.

Section 4: I understand that the main outcome of the discussion is an “advised” number of frames to be used within MFACNet, which is T=2. Is this T the same as used in the comparison section 3? If yes, I would put discussion section before the comparison section, if not, I would state what was the T adopted there.

English quality is overall good.

Author Response

Dear Reviewer,

Thank you for your valuable feedback and constructive comments on our manuscript titled [MFACNet: A Multi-frame Features Aggregating and In-ter-Feature Correlation Framework for Multi-Object Tracking in Satellite Videos || Manuscript ID remotesensing-2917758]. We appreciate the time and effort you have put into reviewing our work. We have carefully considered your comments and suggestions, and we have made the necessary revisions to address them. In attached file, we provide a point-by-point response to each of your comments:

We believe that the revisions we have made have significantly improved the quality and clarity of our manuscript. We have addressed all the concerns raised by the reviewers and have provided additional explanations and evidence where necessary. We hope that these revisions meet your expectations and address any remaining concerns.

Once again, we would like to express our gratitude for your valuable feedback and for helping us improve our manuscript. We believe that your comments have greatly contributed to the overall quality of our work. We look forward to hearing your feedback on the revised version of our manuscript.

Thank you once again for your time and consideration.

Sincerely,

Hu Zhao

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper presented an interesting work for multi-object tracking in satellite videos by integrating multi-frame feature aggregation techniques with inter-feature correlation mechanisms. The results seem pretty good and promising. However, the manuscript needs some revisions addressing the following concerns:

1. Brief quantitative results should be included in the Abstract to demonstrate MFACNet’s superiority.

2. I am not sure if the methods (CenterTrack, FairMOT, TraDes, SORT, OC-SORT, and ByteTrack, line 413) in the manuscript represent the ‘state-of-the-art’ (line 23, line 170). As far as I know, some newer MOT methods such as StrongSORT, BoT-SORT, YOLO v7/v8-based, and Deep OC-SORT have been proposed in the past years. The author should update the latest research in the Introduction and compare them with MFACNet in the Experiments.

3. The author listed and analyzed the algorithms for MOT in satellite videos. How about the object detection and tracking algorithms for common videos (in computer vision) and UAV videos? Can they be adapted for satellite videos? Small-sized objects usually appear in UAV videos as well.

4. The layout of Figure 1 may be confusing as (a) to (c) are instances in satellite videos, and (d) is an instance in remote sensing imagery, but they are in the same row.

5. How about the accuracy of ReID? I did not see the trajectory association results in the Experiments either.

Author Response

Dear Reviewer,

Thank you for your valuable feedback and constructive comments on our manuscript titled [MFACNet: A Multi-frame Features Aggregating and In-ter-Feature Correlation Framework for Multi-Object Tracking in Satellite Videos || Manuscript ID remotesensing-2917758]. We appreciate the time and effort you have put into reviewing our work. We have carefully considered your comments and suggestions, and we have made the necessary revisions to address them. In attached file, we provide a point-by-point response to each of your comments:

We believe that the revisions we have made have significantly improved the quality and clarity of our manuscript. We have addressed all the concerns raised by the reviewers and have provided additional explanations and evidence where necessary. We hope that these revisions meet your expectations and address any remaining concerns.

Once again, we would like to express our gratitude for your valuable feedback and for helping us improve our manuscript. We believe that your comments have greatly contributed to the overall quality of our work. We look forward to hearing your feedback on the revised version of our manuscript.

Thank you once again for your time and consideration.

Sincerely,

Hu Zhao

Author Response File: Author Response.pdf