A Comparative Study of Local Descriptors and Classifiers for Facial Expression Recognition

Round 1

Reviewer 1 Report

The manuscript has been well organized and comprehensively written and the discussion is thorough. I have few suggestions for improvement:

-Please specify the novelty statement of the work at the end of the introduction section as it is currently not clear

-Line 108: Make 'P' in P-bit code italics. Maintain consistency in terminologies throughout the manuscript.

-There are certain typographical errors in the manuscript for example in Line 274 "Based on the results form the previous" should read "Based on the results from the previous"

-In most of the tables, the title of column 1 has not been mentioned. Please add it.

-The RaFD dabase based classifier mentioned in reference [39] performs better than the corresponding proposed model. What could be the possible reason for this difference? Please add this discussion to the section dealing with comparison. Also, the same section can be further elaborated. 

Author Response

See attached document

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper considers the problem of facial expression recognition. This is an important research area and it is within the scope of the journal.  The paper systematically and investigates different combinations of local descriptors and classifiers on different datasets. This is an interesting paper. It is well organized and well written. It is clearly presented.

Abstract: The first time you use the term FER, put the acronym in parentheses after the full term.

Abstract:  Consider changing “Of all the descriptors” to all considered descriptors”. There are other instances in paper where a similar change could be made.

Line 15: Consider rephrasing: “Facial expression recognition (FER) plays a central role in artificial intelligence with several applications”.  E.g. FER has a number of different applications including…

Consider having different subheadings for different types of local descriptors and  classifiers, e.g. 2.1. Local descriptors, 2.2. Classifiers.  In that case the individual local descriptors could be numerated as 2.1.1. Local Binary Patterns” , 2.1.2. Compound Local Binary Patterns, 2.3. Local Directional Patterns, etc.

Line 264: Consider using the term “NB classifier” instead of ”NBC classifier”, to make it more consistent with  figure legends. Or alternatively change the figure legends.

Discuss in more detail and emphasize the main advantages of the proposed methods over the methods (listed in Table 8 Comparison with state-of-the-art methods) that have higher accuracy values.

Author Response

See attached document

Author Response File: Author Response.pdf