FiFoNet: Fine-Grained Target Focusing Network for Object Detection in UAV Images
Round 1
Reviewer 1 Report
Weaknesses:
(-) Experimental evaluation must be improved.
(-) Some improvements are needed in the description of the method.
==== METHOD ====
The authors should first give an overview of their solution before explaining the details.
A novel solution is presented but it is important to better explain the design decisions (e.g. why the solution is designed like that)
The authors should add proof(s) of the properties, theorem or lemmas contained in the paper.
The algorithm(s) should be described clearly using pseudocode.
It is necessary to discuss the complexity of the proposed solution.
==== EXPERIMENTS ====
The experiments should be updated to include some comparison with newer studies.
A statistical analysis should be carried out to demonstrate that the experimental results are significant.
There is not enough discussion of the experimental results.
The experiments have been carried with a few datasets. It is necessary to add more datasets so as to make experiments more convincing.
Some experiment(s) should be added to show that the proposed solution can be used in real applications.
Some additional experiments are required:
- Scalability
- Runtime
- Memory
- Sensitivity analysis
==== REPRODUCIBILITY ====
To ensure reproducibility of the results, the code of the proposed solution should be made public on a website.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
This paper deals with object detection from UAVs and proposes a novel system to process the fine-grained detail of subparts while also removing the background clutter.
Overall I find the paper sufficient to get published. Though I have one point that is a bit not mentioned. They say that their model is run on an RTX-3090 which is a state of the art NVIDIA Graphics card with a lot of processing power and a lot of memory. They do not present their result in practise, i.e. how their model performs on an edge device like a jetson nano that can be flying with the UAV. So this is why I am deciding on a minor revision so as the Authors mention on how their model fit on an edge network. They need to set-up an edge environment with limited resources and compare with the algorithms they already mention.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
The paper proposes a Fine-grained Target Focusing Network (FiFoNet), to improve the performance of object detection in UAV images through aggregating fine-grained objects’ sub-parts with a special focus on foreground target areas. Below I give some of the limitations which I have identified and my suggestions for enriching the paper:
1. It would be nice if authors can provide a table to provide the methods used in previous studies and the drawbacks and advantages of similar studies mentioned in the related work.
2. Add a new section named “Limitation and Discussion” to give a limitation of the proposed method and future research gaps in this field.
3. Rest of the article is written well. Authors should also check the article for typo errors and English grammar.
4. Please check the style and format of references.
5. How does the proposed method work when the image's contrast is low or in rainy weather?
6. To improve the Related Work and Introduction sections authors are recommended to consider this high-quality research work:
Smart Glass System Using Deep Learning for the Blind and Visually Impaired. Electronics.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
