A Fast and Accurate Few-Shot Detector for Objects with Fewer Pixels in Drone Image

Round 1

Reviewer 1 Report

The article is about the a fast and accuracy few-shot detector for objects with fewer pixels in drone image. In general, the article is suitable for publication, but requires corrections first.

From the scientific point of view, the article presents the high scientific level. After taking into account the corrections presented for the authors, it is suitable for publication in the journal Electronics.

The paper should be improved namely in some identified aspects:

- the authors should have some extra effort to make the description more easily understandable, more clear to all the readers and more simple to read as the subject seems a kind of too “complicated”.

- figure 1 is not clear enough (especially descriptions).            

- although the English is OK, the text needs to be simplified.

- the achievements presented in the article should be compared more closely with examples from other countries. 

- there should be more references to existing scientific achievements in the article, because the literature review omits the basic achievements in recent years (examples of achievements were published, among others, in the journal Applied Sciences).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This article investigates object detection from a new perspective--for aerial images with solid experiments and sound results. It adopts ideas from multiple state-of-the-art works to design an efficient one-stage few-shot detector.

The article is overall good, and here are a few suggestions for improvement:

• There are a few grammar issues and typos. For example: in the title, accuracy -> accurate; in the abstract: which lead -> which leads. A few more places misuse "a/an/the". Please review the article and fix any grammar mistakes.
• The overview section (S3) is hard to follow. A suggested way is to organize the section into paragraphs by the motivating problems.
• A few places are missing citations, such as PrROI, the introduction to the dying ReLU issue, etc.
• S7.2 should be put before S7.1.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

As far as I can said, this paper is interesting, well structured and with interesting results that show that the proposed method is relevant.

Author Response

Thank you for your review.