Review 3

[massich]

Reviewer 3 of ICPR 2016 submission 710 (Review 4913)

Comments to the author

This paper is a survey on current state of automatic classification methodologies to identify Diabetic Macular Edema (DME). The major contributions are providing a public benchmarks on a DME dataset. The paper presentation is clear enough to follow, even that it is a survey paper, and number of page is limited.

My comments on this paper are:

• A survey should include some suggestions, or some observations or trend to the issues (in this case, that is DEM identification ). However , no any suggestions on features, or comments on archived performance (Which one are feasible or reasonable to develop applications in daily clinic.)
• Recently, many feature learning method can be applied. In this paper, the authors surveyed on hand-craft designed features (such as, LBP, HOG, Edge, so on). Then it is more interesting if the authors also include results of recent trends of the feature learning topic.