Memory-Based Learning and Fusion Attention for Few-Shot Food Image Generation Method

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper is interesting and describes the research carried out with an appropriate level of detail. The topic is within the scope of the journal.

The experiments are described in detail and the results are compared with the state of the art. In particular the performed experiments demonstrate the validity of the proposed method, outperforming the state of the art. The method is reproducible since the dataset is open.

From this reviewer point of view, the only weak point is in the introduction, where it is hard to understand the solved problem and the current challenges. This reviewer suggest to introduce the problem explaining the general concepts and challenges, avoiding to use acronyms. In the actual version, the introduction can be understand only by experts in this field.

There are a lot of acronyms that are not explained the first time they are used.

Figure 1 is very complicated and there are arrows superimposed on the text. This reviewer suggests to modify this figure.

 

A final remark is about the readability of pseudocodes. This reviewer suggests to insert line numbers and to explain each line in the main text, in order to improve the description of the method.

Comments on the Quality of English Language

English is fine, but please avoid using the short form of verbs.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

1. The idea of ​​the article is very good and certainly relevant in the culinary field. But you won't notice the taste or smell in the pictures. You cannot take this into account from the photo images themselves, so you need to connect some other components and sensors.

2. It might be better to classify ready-made food by-products: bread, cakes, salad, soups, etc. After that, separate the ingredients. Because it is clear to everyone that bread or cakes consist of flour, eggs, etc.

3. Still, I would shy away from standard food, or that offered by world-famous restaurants, including McDonald, etc., and then turn into ingredients.

4. And how do you consider that exactly those ingredients need to be cooked, and baked and how much time should be spent? Cooking recipes are complicated and to make delicious cakes, you must repeatedly add ingredients and then cook or bake...

 

5. You rely only on known datasets, but the ingredients will differ in different countries. Therefore, I recommend referring to a certain country or a certain region.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

In this paper, the authors propose a new methodology for food image generation, named MLA-Diff. First, ingredient and image features are learned and integrated as ingredient-image pairs to generate initial images. After that, image details are refined by using an attention fusion module. The obtained experimental results show the effectivess and efficacy of proposed approach in comparison with those described in literature. The topic is interesting and worth investigating. The manuscript is well-organized and well-written.

 

My remarks are as follows:

In the ‘5. Experimental Results and Analysis’ section, how is the comparability of the obtained results ensured with those from previous similar studies?

In the same section, please include a complexity analysis of the proposed method.

In the ‘Conclusion’ section, the study limitations should be discussed.

 

Technical remarks:

Figure 2: Why is the last ingredient indicator labeled as “B”?

P. 5, the ‘3. Method’ section: Steps 3 – 5, please rewrite the formulas using the Equation editor.

Algorithm 1: Change ‘formula 3’ to ‘formula 2’.

Algorithm 2: Adjust the numbering of the formulas from 4 to 10 to become 3 to 9.

P. 11: How are the reductions (8.64, 0.01, and 1.637) calculated?

In the Data Availability statement, the phrase “& material” should be edited.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf