VSAI: A Multi-View Dataset for Vehicle Detection in Complex Scenarios Using Aerial Images
Round 1
Reviewer 1 Report
The presented article is not a typical research paper, but it contains a multi-view dataset for vehicle detection in complex scenarios based on images taken from the drone. It can be used for ANN trainings, and etc for objects detection.
The paper will be interested for the UAV community, but in my opinion it would be good to show more results of it use, and the best would be to provide codes that can be easily use to teach and learng ANN. With good, validated software it would be something very usefull. Without, it is just the another paper showing possibility of collected dataset.
The paper is written carefully. I just notice one mistake: "arial view" -> should be "aerial view".
I expect precisly state-of-the-art analysis, a very extensive review of similar solutions, and comparison of their pros and cons.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
The subject of the paper is very interesting - Vehicle detection using aerial images with arbitrary orientation is an essential task in remote sensing and computer vision with various applications in traffic management, disaster monitoring, smart city and so on.
The authors briefly and legibly discussed the problems related to object detection in aerial images - many challenges such as image degradation, uneven object intensity, complex background, various scale, and various direction. The authors mainly explain the collection details of the entire VSAI dataset, the basis for category selection (small-vehicle or large-vehicle), and the annotation methods of the VSAI dataset - so called multi-View dataset for vehicle detection in complex Scenarios using Aerial Images (VSAI) for enlightening the object detection research based on drones. Then the authors presented the major characteristics of the proposed dataset VSAI, which consists of multi-view UAV images, objects visibility information and more instances in each image. These properties in comparison to other datasets are sequentially described. The authors presented the results of the evaluation of the proposed dataset VSAI. These issues are the strengths of the article. However, the weaknesses of the article are as follows. The article is not very scientific - the article is of a popularizing nature. Therefore, I propose to add "Materials and Methods" section and rewrite the text in the section "6. Conclusion", because specific conclusions are lacking, while the text is rather a summary; In addition, I noticed minor errors: - Table 1: "COMS" -> "CMOS"; - line 193: "small struck" -> "small truck";- Fig.15: it is very difficult to see the differences between the pictures in the rows; I propose to mark / indicate significant differences in the pictures;
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
The explanations and extensions of the manuscript prepared by the authors are satisfactory