Object Detection-Based Video Compression

Round 1

Reviewer 1 Report

Please refer to the attachment. 

Comments for author File: Comments.docx

Author Response

We would like to thank the reviewer for the thorough review and very helpful comments. The revised manuscript was reviewed and substantially modified by a native speaker.

Please refer to the attached file.

Author Response File: Author Response.docx

Reviewer 2 Report

I have but suggestions to increase the quality of the manuscript only:

(i) Fig. 6 tags are not readable & (ii) the descriptive text included in Fig. 3 and Fig 12 could be represented in the caption for ease of read.

Author Response

We would like to thank the reviewer for the thorough review and very helpful comments. The revised manuscript was reviewed and substantially modified by a native speaker. 

Please refer to the attached file.

Author Response File: Author Response.docx

Reviewer 3 Report

The proposed method is interesting. I only have very minor comments that are not mandatory for a publication:

o page 1, line 43. Something went wrong with the reference.

o page 2, line 74. One mitght delete "in" before "nowadays" and add a comma instead

o page 4, line 148, heading of subsection. "Detetion" should be corrected to "Detection"

o page 6, formula (1). The comments "high QP" and "low QP" seem to be mixed up compared to the description of the algorithm.

o page 7, line 223. One might not end a page with a heading.

o page 8, line 235 "max_obj" instead of "max-obj" ?

o page 10, line 270. "...led to a decrease the average.." sounds odd to me.

 

Author Response

We would like to thank the reviewer for the thorough review and very helpful comments. The revised manuscript was reviewed and substantially modified by a native speaker. 

Please refer to the attached file.

Author Response File: Author Response.docx

Reviewer 4 Report

The submission is very interesting as objects (aka semantics, data interpretation in a mathematical logic sense). I have to admit that I have not followed the most recent development in video compression. However, I am aware of and have been interested in image data interpretation in image analysis.

The paper relies on a firm methodology. The text is easily readable. I learned by reading the submission.

Object detection, if semantics is explored, does not work for general videos. It is application dependent. The experiments should be performed per application area and its main objects. It would not work in general. The submission correctly names the application areas. If the improvement is achieved, e. g., in surveillance videos, it would be great.

My main criticism is that the submission does not provide enough information to duplicate the work, which is the principal requirement in scientific research.  I understand that authors build on top of VCC reference software. A clear description of the addition by the authors is missing. It would be even better if authors provided their implementation either as a running program or its source code.

Tables with numeric results are too detailed. They should have been put on the web and referenced in the paper. I missed a clear and simply expressed lesson from the experiments. The improvements are at the level of small per cent (below 3 per cent, often below one per cent). This outcome could be very much dependent on the implementation. The reader cannot compare the author's results.

My recommendation is to ask for major revision asking the authors to improve the submission in the direction I suggest.

More detailed comments:

• Line 44. Typo. Fullstop should be after [7].
• Line 111: Block Machine Analysis is not described in the submission to the necessary detail. It is the key element of the approach.
• Line 160: Table 1, Table 2: The generalization ability should be tested and described. This is a critical comment.
• Line 279: Typo. Peoples.
• Line 333: I did not understand what is the message from Figures 10 and 11. Does visual comparison suffice?

Author Response

We would like to thank the reviewer for the thorough review and very helpful comments. The revised manuscript was reviewed and substantially modified by a native speaker. 

Please refer to the attached file.

Author Response File: Author Response.docx

Round 2

Reviewer 4 Report

I browsed the updated manuscript. I think it improved and provides executable code and datasets that the reader can repeat the experiment. I recommend that the submission is published. All the best Vasek Hlavac