DSCEH: Dual-Stream Correlation-Enhanced Deep Hashing for Image Retrieval
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
The paper presents a novel approach for improving deep hashing methods for large-scale image retrieval by introducing a dual-stream correlation-enhanced deep hashing framework (DSCEH).
Below are the comments.
1. The process of feature fusion through concatenation is mentioned. Provide details on how fusion improves the retrieval process and any specific techniques used would be helpful.
2. Include more details about the GCN implementation, like how nodes and edges are constructed and the specific GCN architecture or parameters used. This would help in understanding the correlation optimization process better.
3. Briefly explain how VIT creates global representation and its advantages over other methods to improve clarity. Elaborate on the proposed novel approach that leverages both CNN and VIT features. Explain the methodology, how it differs from existing approaches, and the expected improvements in image retrieval performance.
4. Provide specific examples or case studies where GCN has been successfully integrated with hash learning, illustrating the improvements in retrieval tasks. Add more detail on the operations of GCNs, such as the steps involved in feature updates and information propagation. Discussing the underlying mathematical models or algorithms would enhance the technical depth.
5. Elaborate on how the adjacency matrix is constructed, including any techniques or criteria used to define the relationships between nodes. Discuss its impact on the effectiveness of GCN.
6. Provide more technical details on how GCN is integrated into the new framework, including the steps involved in information propagation and relationship reinforcement among image features.
7. Elaborate on the adaptive weighting mechanism used in FGCMH, explaining how the weights are determined and how they contribute to optimizing multi-modal correlations.
8. Add specific results or performance metrics from studies to demonstrate the improvements achieved by GCH, GCNH, and FGCMH. Comparisons with baseline methods would also be beneficial.
9. Providing more details on the optimization process, including convergence criteria and computational complexity, would add depth to the technical discussion.
10. he framework's scalability to large datasets and its computational efficiency during training and retrieval phases should be analysed.
11. The impact of hyper parameters such as the balance constants 𝜎 and 𝛼 on performance needs thorough examination.
12. Testing the framework on diverse and real-world datasets could provide insights into its generalization capabilities and practical effectiveness.
Comments on the Quality of English LanguageGrammar check should be done.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsI am happy with the manner the authors have presented the study. Also, the results seems to be promising. I have few quires which you may use to improve the manuscript.
1. Can you justify why the model performance dropped in MSCOCO and NUSWIDE in a more detailed manner?
2. Suggested to add few visual examples for the positive and negative retrivals in the proposed model.
3. What is the percentage of error in mAP?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsComments to the authors:
1. The motivation and main difference between this paper and the related works should be presented more clearly.
2. New points and advantages of the proposed approach should be shown more clearly
3. Analysis in this paper is quite straightforward, which can limit the contribution of this paper.
4. There are a lot of equations and contents in this paper which are not cited to the sources.
5. Lack of performance comparison with new methods is also drawback of this paper.
6. Almost references in this paper are old, which do not reflect the current research. Therefore, the authors should update the reference list.
7. The conclusion section should be improved with more recommendation and useful designs.
8. There are still several typos in this paper which need to be corrected.
Comments on the Quality of English LanguageModerate editing of English language is required
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe suggested modifications have been incorporated.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe revised manuscript is more solid. This paper can be accepted for the publication.
Comments on the Quality of English LanguageMinor editing of English language is required
