Research on Rapid Recognition of Moving Small Targets by Robotic Arms Based on Attention Mechanisms
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis study focuses on moving small target detection. It extends Yolov8 with an attention mechanism. Some parts should be improved:
1. While EMA is proved to be effective, it is unclear why EMA is used to augment yolov8 rather than other attention mechanisms.
2. The validation is very limited to recognizing one type of object, i.e., apple. The generalization capability of the algorithm is not studied.
3. The related work is far from comprehensive. For example, attention mechanisms have been widely studied in the literature. Relevant works should be included, e.g., Motion-attentive transition for zero-shot video object segmentation, Regional Semantic Contrast and Aggregation for Weakly Supervised Semantic Segmentation, Volumetric memory network for interactive medical image segmentation
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsReview of applsci-2940612: “Research on Rapid Recognition of Moving Small Targets by Robotic Arms Based on Attention Mechanisms”
The subject of the paper is relevant with the topics of the journal. The paper is well structured and gradually introduces the reader to the problem tackled and the aims of the research. It deals with an application of an improved YOLOV8 algorithm, which additionally provides an attention mechanism. The proposed application is made for working with robotic arm control systems. The current work, recognized the gap of long training times and small target detection problems. By reorganizing the calculation of the loss function, the long training times are reduced and small target detection is included.
The references well selected and well organized. Most of them were published the last 3 years, thus taking the state-of-the-art developments into account. The recognition effect of the proposed algorithm was additionally validated, when fruits moved quickly, thus, providing a very good starting point for the model’s robustness. This is true even when blurring effects are considered.
It would increase the quality of the paper if the authors were willing to incorporate the following:
• Line 140, …..The present paper…not this chapter
• Line 141-158, elaborate on the new features incorporated in the new algorithm network.
• Figures 4, 5 and 6, pay attention to increase visibility for better readability.
My proposal to the editors is to accept the paper after minor revisions.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsMain remarks
- The Introduction lacks information about the latest version of YOLOv9. A comment is required as to why the latest version was not used in the research.
- An extended description of the introduced new features is required (Line 140). It is important to emphasize the introduced element of novelty.
- What is the rationale for using the 435i camera (line 262)? Authors' comments are required.
- Why only fruits (line 291)?
- Why were two different hardware platforms used for training and testing (line 322 and line 326)?
- Did the tasks performed by the operating system (e.g. system update) affect the FPS value obtained (Table 2)? If not, how was this achieved?
- The manuscript lacks a justification for choosing Grad-CAM (line 418)?
- The authors mention that there was an improvement in the speed of fruit recognition (lines 426 - 432). Based on Figure 14, it can be assumed that this is an industrial line. What benefits have been achieved in reducing the time needed to recognize fruit?
- The summary lacks an emphasis on the novelty aspect. Description extension is required.
Other minor remarks and recommendations
- Did the algorithm detect mechanical damage to apples? If so, then it is worth mentioning.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe revision has addressed all my concerns.
Author Response
We are grateful for your guidance and patience, and we are committed to maintaining the highest standards of accuracy in our work. Thank you for helping us improve the quality of our submission.