Classification of Ancient Roman Coins by Denomination Using Colour, a Forgotten Feature in Automatic Ancient Coin Analysis (Version 2, Revised)
|Reviewer 1 Nancy Serwint Arizona State University||Reviewer 2 Stephanie Koerner University of Liverpool |||Reviewer 3 Eleni Fragaki Leiden University|
Approved with revisions
Approved with revisions
|Approved with revisions||Reviewer invited|
Ma, Y.; Arandjelović, O. Classification of Ancient Roman Coins by Denomination Using Colour, a Forgotten Feature in Automatic Ancient Coin Analysis. Sci 2020, 2, 37.
Ma Y, Arandjelović O. Classification of Ancient Roman Coins by Denomination Using Colour, a Forgotten Feature in Automatic Ancient Coin Analysis. Sci. 2020; 2(2):37.Chicago/Turabian Style
Ma, Yuanyuan; Arandjelović, Ognjen. 2020. "Classification of Ancient Roman Coins by Denomination Using Colour, a Forgotten Feature in Automatic Ancient Coin Analysis." Sci 2, no. 2: 37.
Article Access Statistics
Arizona State University
Comments on “Classification of Ancient Roman Coins by Denomination Using Colour, a Forgotten Feature in Automatic Ancient Coin Analysis”
Incorrect subject/verb agreement
Erroneous spacing between words
Some phrases (and spelling) are arcane and do not facilitate meaning.
For a reader who is not well-versed in statistics, the article would be confusing.
Authors do not provide enough information for an informed lay person (and possibly a trained numismatist) to understand how the determination of identification of type was reached on the basis of a histogram of color words when an example of the histogram with its range of color words is not provided.
Authors stress that color is only one parameter by which a coin type might be assigned, but one wonders whether dimensions and weight could be considered as well. There is a relatively specific range of diameter for specific coin types, and the same could be said for weight.
Are all the coin images presented in the article shot at the same scale? This is not specified.
It is not at all clear to me (an archaeologist) how the number derived from the formula on page 3 of the article translates into specific words for hue (and its variation) that form the basis of a histogram of color words.
I must say that the premise of the article is an intriguing one and any numismatist (or archaeologist) would be interested in learning more about an approach that permits the variation of specific hue to be accounted for but still allows for a validity of identification of coin types based on color. I found that there was not enough explanation and description of the methodology that was presented in a way that an informed archaeologist could understand, and this detracted from my own professional assessment of the article.
The authors did not comment on whether all the coins that were assessed were derived from the same mint (one would think that this would impact on the variation of the metal content) nor did they address whether commonalities of the dates of issue were taken into consideration.
Response to Reviewer 1Sent on 14 Jul 2020 by Yuanyuan Ma, Ognjen Arandjelovic
University of Liverpool |
Koerner review of sci-746995
Title: Classification of Ancient Roman Coins by Denomination Using Colour, a
Forgotten Feature in Automatic Ancient Coin Analysis
by Yuanyuan Ma, Ognjen Arandjelovic *
for special issue if SCI - Machine Learning and Vision for Cultural Heritage
This article is a strong mix between potential and difficulties in terms of clarity of aims - approaches - findings and conclusions.
The themes and purposes of this article are extremely interesting and have considerable bearing upon the novel aims of the Special Issue
e.g., developments in computer "machine" based "learning" (knowledge acquisition, production, distribution, use) and research in cultural heritage
in these connections - it bears noting extraordinary possible comparisons between themes of the article - and the Special Issue and key themes at the heart of connections in early modern times between innovations in the production, use and distribution of printed texts - and, especially, engraved images and innovations in the roles of the extremely ancient field of numismatics in the dynamics of antiquarian and historical study of the past - with attention to "cultural heritages"
it also bears noting the potential of the extent to which the emphasis the article places on "colour" as a significant line numismatic evidence parallels comparable arguments at the heart of the roles the latter innovations played in deep and far reaching - even 'revolutionary' change in historical reasoning and humanities centring on material culture.
There are, however, difficulties in terms of clarity of aims - approaches - findings and conclusions. These are further complicated by the extent to which key sections lack general orienting details needed to bring relevant points into relief.
Addressing this problem - along with general writing problems (complex sentences, grammar etc) might help bring valuable insights and findings into relief -- and with attention to relevance for the aims of the Special Issue.
Response to Reviewer 2Sent on 14 Jul 2020 by Yuanyuan Ma, Ognjen Arandjelovic
The suggested methods are definitely interesting both for computer scientists and for archaeologist, but at the present state they can only be understood by scholars familiar with these particular issues. They could have been described in a more extensive and explicative way in order to be as accessible as possible to specialists in numismatics, and maybe also to the general public. For this purpose, terminology needs to be elucidated and processes need to be detailed and formulated in simple terms. It should also be made clear for non-specialists why those tools are preferable to previous proposals and allow to obtain better results.
Moreover, the proposed application of these color-based representations does not correspond to the best way to take advantage of them. Their use for classification purposes raises many questions, as colors are a changeable factor which cannot be considered as a permanent and strictly defined characteristic of a specific category of coins. The described procedures could be rather used for the depiction of coins in scientific papers, but also in exhibitions, catalogues, books for a wider readership etc. They can also be included as additional indicators in databases, in order to study and compare specific objects in relation to their history and archaeological context. The aims of the paper should therefore be reviewed so as to provide a more relevant connection with the needs and particularties of numismatics. The related parts of the text should be thoroughly rewritten.