All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
The authors have addressed the reviewers' comments and the paper looks good now. It can be accepted.
[# PeerJ Staff Note - this decision was reviewed and approved by Rong Qu, a PeerJ Computer Science Section Editor covering this Section #]
A minor revision is needed before potential acceptance. Thanks.
The authors answered all comments
The authors answered all comments
done
none
No comment
No comment
No comment
The quality of the revised manuscript is improved, the authors need to deeply revised paper. After addressing this minor comment, I recommend accept decision for this paper.
1) The authors need to revise the sentence in line 466 and avoid the repetition of words.
Previous version
Updated version
The reviewers still have some lingering comments that need to be addressed. Please note that I do not expect you to cite any recommended references unless it is essential. Thanks.
• The novelty of this paper is not clear. The difference between present work and previous Works should be highlighted.
• The author needs to change the abstract and focus more on the problem domain. Before the paper's contributions, the author could precisely include the need to develop the proposed method.
• The author could better explain how “Related works” is actually related to the current study. It is unclear to the reader how the manuscript is similar to or differs from these related works. https://doi.org/10.3390/math10050733; https://doi.org/10.1016/j.imavis.2021.104229
• The Experiment part of the paper is good, however, the authors should include some. Image examples in order to make the experiments and their results more understandable.
The conclusion of the proposed work is discussed with limited content and the
achieved performance value is more efficient when compared to the existing methods.
The conclusion must discuss in detail the limitations of current knowledge, and the
overall importance of the work.
The English should be polished
The quality of the revised manuscript is improved, however there are some minor issues that need to be addressed.
1. In the experimental section, authors report comparison performance of proposed model and SOTA techniques but the accuracy of these techniques on UCF-Crime dataset not UCF Crime2Local. Author should perform ablation studies of these techniques or remove these results from Table.3.
2. The authors need to revise the sentence in line 466 and avoid the repetition of words.
3. The authors claim that “We have implemented our implementation using NVIDIA Jetson Nano” but they not reported time complexity of the model. The authors need to add time complexity and qualitative analysis of the model. For ease see of the authors see the link [1].
Three detailed reports have been received. Please revise the paper according to the comments. A detailed response letter is needed. Please note that you should not cite any recommended references unless it is crucial.
[# PeerJ Staff Note: It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors are in agreement that they are relevant and useful #]
• The author needs to change the abstract and focus more on problem domain. Before the paper contributions, the author could precisely include the need of developing the proposed method.
• The author could better explain how “Related works” is actually related to the current study. It is not clear to the reader how the manuscript is similar to or differs from these related works.
• What are all the advantages and significance of the proposed method? Mention it clearly in the proposed method.
• How did the authors apply the Augmentation technique?
• In case you filliped images, is that enough to create diversity in the data?
• In reality, the images contain specific characteristics or features for each class. How is flipping or rotating the images will create such diversity?
• The biggest concern is regarding trying to grasp what the output of ROI localization is in respect to the object detection model to identify molars.. Is this separate from the molar segmentation image or is it included in this image? And how exactly do these feed into the CNN? The AlexNet CNN requires a fixed input image size
• While lots of segmentation networks like U-Net are available, why don’t use optimization-based segmentation? What are the motivation and advantages? How about changing to segmentation networks?
• The conclusion part should indicate the implications of the experimental evaluation and include some obtained values to point out the superiority clearly.
• Some recent works should be added such: https://doi.org/10.1016/j.imavis.2021.104229; https://doi.org/10.3390/math10050733
• Figures not clear please show it in high resolution
• Without providing cross-validation, it is very difficult to check the performance of their model. I strongly recommend the authors add this analysis and compare it with other models. And References must be updated
• How many iterations are taken while finding fitness values in optimization algorithms?
• What are the parameter settings involved in the validation of the proposed method?
• The Experiment part of the paper is good, however, the authors should include some. Image examples in order to the experiments and their results be more understandable.
• It will be better if you mentioned the future work and the proposed model limitations at the end of the conclusion.
no comment
no comment
no comment
The authors proposed a Temporal Based Anomaly Recognizer (TAR) framework for anomaly recognition. To make anomaly detection more resource efficient, the authors took the MobileNet-V2 as the backbone and used residual shift operation to incorporate the temporal features. Then, they performed two class and multi-class anomaly recognition on the surveillance video and conducted experiments on the UCF Crime2Local dataset. The results are fine.
In the experimental section, the authors compare the accuracy, the model size, and the amount of computation of different methods in detail. It would be better if authors provide the runtime or the process frame rate. Besides, the authors lacked an analysis on the effect of the partial temple shift operation.
The experiments demonstrate the performance and effectiveness of the proposed framework. However, it would be better if there is the ablation study about each component in the proposed model.
1. The format of the references does not seem to be correct.
2. Something is missing in line 173.
3. In the related work section, there is a lack of comparison between anomaly detection and anomaly recognition. Please cite and discuss the video anomaly detection algorithms, e.g.:
[1]Liu W, Luo W, Lian D, et al. Future frame prediction for anomaly detection–a new baseline. Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 6536-6545.
[2]Chang Y, Tu Z, Xie W, et al. Clustering driven deep autoencoder for video anomaly detection. European Conference on Computer Vision. 2020: 329-345.
[3]Chang Y, Tu Z, Luo B, and Yuan J. Video Anomaly Detection with Spatio-Temporal Dissociation. Pattern Recognition, vol.122, pp.108213:1-12, 2022.
[4]Pang G, Yan C, Shen C, et al. Self-trained deep ordinal regression for end-to-end video anomaly detection. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020: 12173-12182.
In this article, the authors proposed a system for anomaly detection and recognition in a surveillance environment. They perform two sets of experiments to achieve a real-time surveillance system. The presented idea is worthy therefore I recommend some major suggestions to enhance the quality of the paper.
The introduction section doesn’t follow the standard rules of manuscript writing. The authors need to revise the introduction section by incorporating the following points:
• Motivation: general (why the problem is important)
• Problem statement
• Current challenges
The authors claim in the third contribution that “Our proposed framework TAR performs anomaly localization” but there are no experimental results in this aspect. The authors should either revised the contribution or add experimental results.
In the “TEMPORAL ANOMALY RECOGNIZER” section, the authors need to revise this section's lack of temporal information relationship learning details.
In the overall paper the captions of figures are confusing, and the authors should include complete description of figures and Tables.
The experimental section is very weak in anomaly recognition and lacks the confusion matrix and class-wise accuracy see the links [1, 2].
The experimental section is very weak in anomaly recognition and lacks the confusion matrix and class-wise accuracy see the links [1, 2]. The authors provide only a comparison table.3 without detail. Furthermore, in Table.1 authors only provided the ResNet50 and MobileNetV2 parameters, and TAR parameters were not added to the final model TAR+ MobileNetV2 and TAR+ ResNet50. The authors should add full parameters of the TAR+ MobileNetV2 and TAR+ ResNet50.
1) https://www.sciencedirect.com/science/article/pii/S0167739X21004295
2) https://arxiv.org/ftp/arxiv/papers/2101/2101.01073.pdf
In the conclusion section, the authors should discuss the limitations of the proposed method and their possible solutions in future work.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.