Next Article in Journal
A Self-Tuned Method for Impedance-Matching of Planar-Loop Resonators in Conformable Wearables
Next Article in Special Issue
Sequential Clique Optimization for Unsupervised and Weakly Supervised Video Object Segmentation
Previous Article in Journal
A Hybrid Method for Keystroke Biometric User Identification
Previous Article in Special Issue
Industry-Fit AI Usage for Crack Detection in Ground Steel
 
 
Article
Peer-Review Record

LNFCOS: Efficient Object Detection through Deep Learning Based on LNblock

Electronics 2022, 11(17), 2783; https://doi.org/10.3390/electronics11172783
by Beomyeon Hwang 1, Sanghun Lee 2,* and Hyunho Han 3
Reviewer 1:
Electronics 2022, 11(17), 2783; https://doi.org/10.3390/electronics11172783
Submission received: 20 July 2022 / Revised: 26 August 2022 / Accepted: 2 September 2022 / Published: 4 September 2022
(This article belongs to the Collection Computer Vision and Pattern Recognition Techniques)

Round 1

Reviewer 1 Report

The authors propose an anchor-free deep-learning method based on fully convolutional one-stage object detection with the aim of achieving an optimal trade-off between computational cost and accuracy.

The results presented by the authors clearly show that the computational cost of the proposed method is contained, whereas the accuracy is improved.

I think the paper is well-structured and easy to read.

I point out a probable typo at line 87-89.

Author Response

Dear Reviewer

First of all, thank you for reviewing our paper.

We will answer your questions.

Q.1. I point out a probable typo in lines 87-89.

A.1. I couldn't get rid of unnecessary phrases in the process of editing after proofreading. Accordingly, we removed paragraphs line 87-89.

Thank you for your review.

best regards authors

Reviewer 2 Report

1.The authors should explain the mAP (%) value, in Table 2, that is negative: -74.6%
2.Also in table 3,  LNFCOS(our) values are sometime not so good that other newtorks (e.g. One-stage, column AP50 (%), 56.0% < 57.9%) The authors must explain there results.
3. In “conclusions” section, a more emphasis on original contribution would be desirable.

The range of the paper is 95% from 100% .

Author Response

친애하는 리뷰어

먼저 저희 논문을 검토해 주셔서 감사합니다.

귀하의 질문에 기꺼이 답변해 드리겠습니다.

Q1. 저자는 표 2에서 음수인 mAP(%) 값을 설명해야 합니다. -74.6%

A.1. Latex 작성 시 "&" 기호 누락을 확인하여 표기법을 수정하였습니다.

Q.2. 표 3에서 LNFCOS(우리) 값은 때때로 다른 네트워크(예: One-stage, 열 AP50(%), 56.0% < 57.9%)에서 결과를 설명해야 할 정도로 좋지 않습니다.

A.2. 비교적 쉬운 객체에 약하다고 생각하지만, 제안하는 방법을 통해 어려운 객체(AP 75 ~ AP95) 간의 탐지 성능이 향상되는 것을 확인하였다. 그 부분에 위의 내용이 추가되었습니다.

Q.3. "결론" 섹션에서는 독창적인 기여를 더 강조하는 것이 바람직할 것입니다.
종이의 범위는 100%에서 95%입니다.

A.3. 말씀하신대로 결론부분 수정했습니다.

검토해 주셔서 감사합니다.

안부 작가

Author Response File: Author Response.pdf

Back to TopTop