Hamilton Kerr Institute: Difference between revisions

From Computer Laboratory Group Design Projects
Jump to navigationJump to search
No edit summary
No edit summary
 
(One intermediate revision by the same user not shown)
Line 1: Line 1:
Contacts:
Contacts:
Camille Polkownik <cp574@cam.ac.uk>


Spike Bucklow <sb10029@cam.ac.uk>
Spike Bucklow <sb10029@cam.ac.uk>


2018 suggestion:
2021


[[Pigment Analysis]]
Spike says: "Digital stuff is now a very small part of my work - I have a Leverhulme project doing machine learning with DAMTP - but I'm not sure what I could come up with for your students. And we don't have any interns like Camille this year because of covid so I can't  ask anyone else to come up with a project either. Sorry."


===Correspondence===
2019


The project is to build an interactive and intuitive diagnostic tool that would enable conservators to identify pigments through their microscopic characteristics. The tool would work as a deconstructed flowchart: the conservator can start with whichever characteristic he/she wants. The more characteristics you enter, the few options are presented, narrowing the search. This is more intuitive and allows more freedom as the analysis is not always straightforward. All the information on pigment characteristics has already been typed in tables, allowing copy/paste to make the job easier and faster. As of now, there are 84 pigments in the list but more are being added. Total number should be around 100 (provisional). This educational tool would be online and in free-access. If successful and accurate, the website would be supported by the IT staff from the Fitzwilliam Museum and would be hosted at the Hamilton Kerr Institute in Whittlesford.
Could follow up exchange with Kasia Targonska-Hadzibabic <kt441@cam.ac.uk>, who suggested doing some content analysis of paintings. This could apply methods from recent PhD or Leonardo Impett.


Feedback:
2018


I would be inclined to add a slightly more ambitious technical challenge, so that some members of the team can look at extending the capability beyond your suggestion of a manual search. This might use automated image analysis techniques from the OpenCV computer vision libraries, together with a simple classifier, to assist the user with some artificial intelligence support. The key would be that expert users should be freed up to concentrate on the “not straightforward” aspects of the work, with automated decision support assisting with the mundane parts.
Project with Camille Polkownik <cp574@cam.ac.uk> (now left Cambridge)


Response:
[[Pigment Analysis]]
 
Dear Alan, Thank you for your email. This is very interesting because Spike and I were talking about AI and how to apply it for a diagnostic tool! I just did not want to put it in my proposal at first and present a project that would be too complicated or overwhelming. Also, I'm afraid it just is not my area of expertise, but I am open to suggestions that might make the tool more efficient. I have added your suggestions to the proposal below. Thank you very much for your help and helpful suggestions. I am very excited by the project, I hope it works out! Kindest regards, Camille
 
 
The interactive and intuitive diagnostic tool that would enable conservators to identify pigments through their microscopic characteristics. The tool would work as a deconstructed flowchart: the conservator can start with whichever characteristic he/she wants. The more characteristics you enter, the few options are presented, narrowing the search. This is more intuitive and allows more freedom as the analysis is not always straightforward. The use of automated image analysis techniques from the OpenCV computer vision libraries, together with a simple classifier to assist the user with some artificial intelligence support is a strong suggestion. The key would be that expert users should be freed up to concentrate on the “not straightforward” aspects of the work, with automated decision support assisting with the mundane parts. All the information on pigment characteristics has already been typed in tables, allowing copy/paste to make the job easier and faster. As of now, there are 84 pigments in the list but more are being added. Total number should be around 100-125 (provisional). This educational tool would be online and in free-access. If successful and accurate, the website would be supported by the IT staff from the Fitzwilliam Museum and would be hosted at the Hamilton Kerr Institute in Whittlesford.





Latest revision as of 12:30, 14 October 2020

Contacts:

Spike Bucklow <sb10029@cam.ac.uk>

2021

Spike says: "Digital stuff is now a very small part of my work - I have a Leverhulme project doing machine learning with DAMTP - but I'm not sure what I could come up with for your students. And we don't have any interns like Camille this year because of covid so I can't ask anyone else to come up with a project either. Sorry."

2019

Could follow up exchange with Kasia Targonska-Hadzibabic <kt441@cam.ac.uk>, who suggested doing some content analysis of paintings. This could apply methods from recent PhD or Leonardo Impett.

2018

Project with Camille Polkownik <cp574@cam.ac.uk> (now left Cambridge)

Pigment Analysis


2017

Proposal for a student project Build an Interactive and Intuitive Tool to Assist Art Conservators in Pigment Analysis

The Project

Build an interactive tool (website or app) that would enable conservators to identify pigments through their microscopic characteristics. The reason why we would like to make this a student project is that this educational tool would be in online and in free-access . It would benefit students and recent graduates in conservation as well as older conservators who use polarised light microscopy every now and then and need the support. It also offers the IT student a collaboration with professionals and most of all, a multidisciplinary experience .

How are pigments analysed?

As of 2017, various methods are available to analyse pigments. To cite a few, there is: x-ray fluorescence (XRF), Macro X-ray fluorescence (MA-XRF), SEM-EDX (Scanning Electron Microscopy with Energy Dispersive X-Ray analysis), FTIR (Fourier transform infrared spectroscopy). The major drawback with all these methods is the price: they are expensive to acquire and need specific knowledge of the machine and its limitations. Another method to analyse pigment exist: polarised light microscopy (PLM). It has the potential of solving problems confidently and quickly , as the sample set-up and analysis can be done under 15-20 minutes. Plus, the investment is minimal , one needs a hot plate and glass slides for the mounting of samples and a good microscope.

What is currently available to support pigment analysis? - Flowcharts (paper): very rigid . If one characteristic is missed then it is impossible to go on. - McCrone Atlas ( www.mccroneatlas.com ): not adapted for historical pigments. - Pigment Compendium: dictionary, not an analytical tool . (Eastaugh, N., Walsh, V., Chaplin, T., and Siddall, R., 2004. The Pigment Compendium: Dictionary of Historical Pigments , Oxford: Elsevier Butterworth-Heinemann)

Who would use the interactive tool?

The survey done in July 2017 received 226 responses, mostly from conservators interested in the project and sharing they would like to have access to such a tool. A few even offered to test it for us before the launch. Interviews with conservation students and emerging and settled conservators alike showed that such a tool would be welcome and would provide support which would itself bring more confidence.

How is the tool different and useful? The tool would work as a deconstructed flowchart : the conservator can start with whichever characteristic he/she wants. The more characteristics you enter, the few options are presented, narrowing the search. This is more intuitive and allows more freedom as the analysis is not always straightforward.

Additional information - The content is being written up by myself, based on available information in the Pigment Compendium and information from flowcharts. - The website/app would be supported by the IT staff from the Fitzwilliam Museum and would be hosted at the Hamilton Kerr Institute in Whittlesford. - The design, amongst other things, would be up to the student and every aspect would be discussed to produce the most accurate project.