첫 리뷰가 왔다!!! 아래와 같은 이메일과 4명의 리뷰어의 평과와 comments 그리고 논문에 주석을 통해 수정을 제안해주셨다.
Dear Mr. Jinkyu Lee,
The review of your paper, "Real-Time Downward View Generation of a Vehicle Using Around View Monitor System," T-ITS-18-11-1077, has been completed. The reviewers’ comments and those of the Associate Editor are copied below. Based on these comments and recommendation of the Associate Editor your paper is not ready for publication in its present form, A properly revised version that takes care of the concerns and drawbacks pointed out by the reviewers and Associate Editor is potentially publishable.
Therefore, I suggest that you revise your paper along the lines described by the reviewers and resubmit the paper. Please include a description on how you took into account the reviewers' comments in preparing your revision. We would then hope to determine a publication decision soon thereafter. Please note that if your revision is not submitted within the next 3 months (90 days), your paper will be treated as a new paper.
You may want to resubmit your paper as a REGULAR PAPER (suggested length: 10 Transactions pages, authors' biographies included). The Associate Editor will determine whether your paper is best suited to be considered as a Regular Paper or shortened to a Short Paper.
Thank you for submitting your manuscript to the Transactions on Intelligent Transportation Systems.
Sincerely,
Azim Eskandarian
Editor-in-Chief
Transactions on Intelligent Transportation Systems
Reviewer: 1
- Recommendation: Accept With Minor Changes
- Summary of Evaluation: Good
- Organization: 4
- Clarity: 4
- Length: 4
- References: 2
- Correctness: 4
- Significance: 4
- Originality: 3
- Contribution: 4
What are the contributions of this paper?:
This paper proposes a method to generate the downward image of current vehicle location using AVM. The system includes feature tracking, obstacle filtering, and downward view generation. The Shi-Tomasi corner is detected and then tracked by a KLT tracker. Obstacles are filtered by using a histogram based on the movements of the features. RANSAC is used to remove the outliers. The result indicates that downward images are generated accurately.
What are some ways in which the paper could be improved? Please supply any additional important references that you feel the author omitted which should be noted in the paper.:
The paper is basically well written. The downward view image may lack the information that is far away from the vehicle, but they are useful for certain cases like in the parking lot or charging pad.
The ‘Related Work’ part seems insufficient or incomplete. It has a single sub-section ‘A. Around View Monitor (AVM)’. It looks like the authors plan to add more sub-sections? Some papers use the terms ‘bird-eye view’ or similar. The existing methods (e.g., bird-eye view generation) are probably not thoroughly covered/compared with in the ‘Related Work’.
Reviewer: 2
- Recommendation: Prepare A Major Revision For A New Review
- Summary of Evaluation: Fair
- Organization: 4
- Clarity: 2
- Length: 4
- References: 4
- Correctness: 3
- Significance: 3
- Originality: 3
- Attachments: 3
- If Survey Coverage: 3
- Contribution: 3
What are the contributions of this paper?:
The paper shows how to make an all-around view of the vehicle, including the area under the vehicle.
What are some ways in which the paper could be improved? Please supply any additional important references that you feel the author omitted which should be noted in the paper.:
The English language should be improved - eg checked by native English speaker and, some paragraphs - especially the description of the algorithm and the test results - are very hard to understand. Some sections could be supported through figures.
A clear ADAS use case for including the area under the vehicle is missing: the driver has very limited possibilities to avoid obstacles under the vehicles in the track of the wheels; obstacles in the area which is not in the wheels track are not relevant, except for accurate positioning over a point under the vehicle - if there would be such a use case. The paper gives no information on the need of overlapping of successive images, and at the start of the paper it is not clear where the cameras are installed.
Reviewer: 3
- Recommendation: Reject
- Comments:
The quality of this paper is far below the acceptance bar. There are many issues with the novelty, algorithms, as well as the comparison with benchmark. It cannot be accepted. - Summary of Evaluation: Poor
Reviewer: 4
- Recommendation: Accept With Minor Changes
- Comments: Read again to check for minor typos.
- Summary of Evaluation: Good
- Organization: 4
- Clarity: 4
- Length: 3
- References: 2
- Correctness: 4
- Significance: 3
- Originality: 3
- Contribution: 4
What are the contributions of this paper?:
The paper describes a three steps method to generate the downward image of current vehicle location using AVM image.
The main advantages of the proposed method are that the processing is done in real-time, thus helping the driver in manoeuvers like parking and allowing safe driving, and that it is based on commercial AVM systems, so it does not require expensive sensors setup.
The steps of the proposed system are clearly explained, and the selected approaches motivated
What are some ways in which the paper could be improved? Please supply any additional important references that you feel the author omitted which should be noted in the paper.:
Even if the proposed method is well motivated and clearly described, related work section is a little poor, since it only describes how AVM image is created from a fisheye camera. In the introduction, some works using AVM image for detecting lanes or parking lots are cited but there isn't any reference on previous works creating or using downward image. From this point of view, the reference list is incomplete, since it is not clear if other studies focused on the creation of a downward image with different approaches, thus making difficult to evaluate the novelty of the proposed method.
Revision Plan
1. AVM과 유사한 시스템의 관련어 검색을 통해 더 이전 시스템에 대한 조사 및 레퍼런스가 필요하다. 이를 통해 본 시스템의 novelity를 밝혀한다.
2. related work를 추가하기
3. Use Case와 Contribution을 납득가능하게 잘 설명하기
4. 전반적인 표현 및 설명을 다시 한 번 수정하고 보안하기