Discussion - Member to Member Sales - Research Center
Discussion - Member to Member Sales - Research Center
Since there are so many variables that have to do with correct stamp identification, it is WAY PAST TIME ffor a software engineer to write code that when a stamp is scanned, the software could find things like hidden marks, or long ears and also very importantly the true color of the stamp.
For example I have a lot of air mail stamps C 23 and they all look like they are the same color blue. But a proper piece of AI Generated software could distinguish between blue and ultramarine.
Also thie AI progra could quickly check for signs that a stamp is a counterfeit.
JUST A wednesday aftenoon thought.
1 Member
likes this post.
Login to Like.
How about detection of flat press printed vs rortay press printed stamps? In my 67 years of stamp collecting a lot of collectors get these stamps mixed up alot!
2 Members
like this post.
Login to Like.
I'll see what I can do.
i can code a little, but my Dad HAS been teaching me...
It has been on my mind for a while now too...
it would be my first big code project.
Still pondering if it should be free...
It would take me a year, and if it was free, I'd have to put ads,right?
Thoughts!
Edit: I KNOW there are better developers here on StampoRama than me.
Login to Like
this post
Try "Google Search by Image". https://keywordtool.io/image-search
It's not bad and it's free!
2 Members
like this post.
Login to Like.
Add the ability to read watermarks with an image of the watermark!
Login to Like
this post
Note taken, sir!
Login to Like
this post
My short answer is no it is never too late. However AI software is not infallible as it is created/invented by humans. Batteries die and the power goes out. More importantly AI should never replace the basic fundamentals of identification.
Jeremy
Login to Like
this post
Sarge
I agree, but there are some older stamp collectors with limited abilities.
Login to Like
this post
1899,
I hear you loud and clear as you well know I'm on the bottom rung of the old farts ladder. It was last year that I got my first pair of prescribed glasses and completely agree that an AI program would prove very useful for that intended purpose. But I fear that the younger new comers to the hobby would become solely reliant on AI for all their philatelic means and know none of the basics. For that reason and that one alone is why I'm teaching my son the basics as I learned them.
Jeremy
2 Members
like this post.
Login to Like.
I second that!
Every promt from Gemini,Perplexity Pro, ChatGOT etc; has been wrong/lies!
It makes up stuff, and mashes together infi from three sites:
mysticstamp,amazon&ebay,sometimes a forum
use it with a grain of salt...
Except for coding, its great for vibe coders!
Login to Like
this post
1899y friend Mike that does expertising of the Washington Franklin series, told me 1 simple way to see if a stamp is flat plate printed is to look at the back of the stamp. Because sheets of stamps coming off of the flat plate printer were stacked 1 on to of the other there is usually (But not always) some ink transfer on the back of the stamp fro the top sheet.
One problem would be if the first sheet of flat plate printed stamps were on the bottom of the pile there would be no ink transferred to the back of that entire sheet of 400 stamps.
Login to Like
this post
...and while we are at it, tell us the perf. count too!
Login to Like
this post
I did learn on the Scott catalog call that Clive Levinson of Album Easy is working on color id app.
Login to Like
this post
I can see it as an almost impossible task bearing in mind the need to have a huge database of illustrations of the worlds stamps plus variations, errors etc. Then you would have to account for the same stamp but with different watermarks and orientations thereof. Also in regard to perfs there would need to be a way of identifying the various types plus chambons and also imperfs. This would also require an input image showing a measuring scale against the stamp that is to be identified.
Even reverse image sites often do not correctly identify uploaded stamp images, let alone account for cinderellas and forgeries and all the fake stamps that abound.
Our best bet, in my opinion, is a good catalogue, an accurate gauge and a watermark tray. Anything else would be a bonus, though a country identifier would be a boon to our hobby, especially with foreign inscriptions such as the Indian States.
1 Member
likes this post.
Login to Like.
I certainly would appreciate a color database, though every monitor is different.
-Ari
Login to Like
this post
Every sensor used for scanning is also different. The variances in nomenclature used by different stamp catalog publishers also contribute to the issue. Also the issues of brightness, tint, shadows in scans from light sources when using cameras, mixtures of source wavelengths, and even variances in how different humans see color all contribute to the problem.
1 Member
likes this post.
Login to Like.
Don’t forget to throw in cancellations, ink oxidation, paper changes with time, …!
1 Member
likes this post.
Login to Like.
It would be great if ALL the manufacturers of Colour Charts had the same definitions of each colour and even if the catalogue manufacturers used the same descriptions.
Even Stanley Gibbons have issued colour charts that have been "updated" by changing colour descriptions and definitions.
Login to Like
this post
I have just started a new thread called "ChromaMate - colour matching software".
I had initially started to post here, but thought that it might be useful to start a new thread instead.
Clive
1 Member
likes this post.
Login to Like.
06:08:58pm
There are a few coin identification piees of software that are for your phone.
Since there are so many variables that have to do with correct stamp identification, it is WAY PAST TIME ffor a software engineer to write code that when a stamp is scanned, the software could find things like hidden marks, or long ears and also very importantly the true color of the stamp.
For example I have a lot of air mail stamps C 23 and they all look like they are the same color blue. But a proper piece of AI Generated software could distinguish between blue and ultramarine.
Also thie AI progra could quickly check for signs that a stamp is a counterfeit.
JUST A wednesday aftenoon thought.
1 Member
likes this post.
Login to Like.
re: It is way past time for some software engineer to write a stamp identification piece of software.
How about detection of flat press printed vs rortay press printed stamps? In my 67 years of stamp collecting a lot of collectors get these stamps mixed up alot!
2 Members
like this post.
Login to Like.
Approvals
re: It is way past time for some software engineer to write a stamp identification piece of software.
I'll see what I can do.
i can code a little, but my Dad HAS been teaching me...
It has been on my mind for a while now too...
it would be my first big code project.
Still pondering if it should be free...
It would take me a year, and if it was free, I'd have to put ads,right?
Thoughts!
Edit: I KNOW there are better developers here on StampoRama than me.
Login to Like
this post
re: It is way past time for some software engineer to write a stamp identification piece of software.
Try "Google Search by Image". https://keywordtool.io/image-search
It's not bad and it's free!
2 Members
like this post.
Login to Like.
re: It is way past time for some software engineer to write a stamp identification piece of software.
Add the ability to read watermarks with an image of the watermark!
Login to Like
this post
Approvals
re: It is way past time for some software engineer to write a stamp identification piece of software.
Note taken, sir!
Login to Like
this post
10:51:36am
re: It is way past time for some software engineer to write a stamp identification piece of software.
My short answer is no it is never too late. However AI software is not infallible as it is created/invented by humans. Batteries die and the power goes out. More importantly AI should never replace the basic fundamentals of identification.
Jeremy
Login to Like
this post
re: It is way past time for some software engineer to write a stamp identification piece of software.
Sarge
I agree, but there are some older stamp collectors with limited abilities.
Login to Like
this post
01:23:53pm
re: It is way past time for some software engineer to write a stamp identification piece of software.
1899,
I hear you loud and clear as you well know I'm on the bottom rung of the old farts ladder. It was last year that I got my first pair of prescribed glasses and completely agree that an AI program would prove very useful for that intended purpose. But I fear that the younger new comers to the hobby would become solely reliant on AI for all their philatelic means and know none of the basics. For that reason and that one alone is why I'm teaching my son the basics as I learned them.
Jeremy
2 Members
like this post.
Login to Like.
Approvals
re: It is way past time for some software engineer to write a stamp identification piece of software.
I second that!
Every promt from Gemini,Perplexity Pro, ChatGOT etc; has been wrong/lies!
It makes up stuff, and mashes together infi from three sites:
mysticstamp,amazon&ebay,sometimes a forum
use it with a grain of salt...
Except for coding, its great for vibe coders!
Login to Like
this post
09:29:09am
re: It is way past time for some software engineer to write a stamp identification piece of software.
1899y friend Mike that does expertising of the Washington Franklin series, told me 1 simple way to see if a stamp is flat plate printed is to look at the back of the stamp. Because sheets of stamps coming off of the flat plate printer were stacked 1 on to of the other there is usually (But not always) some ink transfer on the back of the stamp fro the top sheet.
One problem would be if the first sheet of flat plate printed stamps were on the bottom of the pile there would be no ink transferred to the back of that entire sheet of 400 stamps.
Login to Like
this post
re: It is way past time for some software engineer to write a stamp identification piece of software.
...and while we are at it, tell us the perf. count too!
Login to Like
this post
re: It is way past time for some software engineer to write a stamp identification piece of software.
I did learn on the Scott catalog call that Clive Levinson of Album Easy is working on color id app.
Login to Like
this post
re: It is way past time for some software engineer to write a stamp identification piece of software.
I can see it as an almost impossible task bearing in mind the need to have a huge database of illustrations of the worlds stamps plus variations, errors etc. Then you would have to account for the same stamp but with different watermarks and orientations thereof. Also in regard to perfs there would need to be a way of identifying the various types plus chambons and also imperfs. This would also require an input image showing a measuring scale against the stamp that is to be identified.
Even reverse image sites often do not correctly identify uploaded stamp images, let alone account for cinderellas and forgeries and all the fake stamps that abound.
Our best bet, in my opinion, is a good catalogue, an accurate gauge and a watermark tray. Anything else would be a bonus, though a country identifier would be a boon to our hobby, especially with foreign inscriptions such as the Indian States.
1 Member
likes this post.
Login to Like.
Approvals
re: It is way past time for some software engineer to write a stamp identification piece of software.
I certainly would appreciate a color database, though every monitor is different.
-Ari
Login to Like
this post
re: It is way past time for some software engineer to write a stamp identification piece of software.
Every sensor used for scanning is also different. The variances in nomenclature used by different stamp catalog publishers also contribute to the issue. Also the issues of brightness, tint, shadows in scans from light sources when using cameras, mixtures of source wavelengths, and even variances in how different humans see color all contribute to the problem.
1 Member
likes this post.
Login to Like.
re: It is way past time for some software engineer to write a stamp identification piece of software.
Don’t forget to throw in cancellations, ink oxidation, paper changes with time, …!
1 Member
likes this post.
Login to Like.
09:40:23am
re: It is way past time for some software engineer to write a stamp identification piece of software.
It would be great if ALL the manufacturers of Colour Charts had the same definitions of each colour and even if the catalogue manufacturers used the same descriptions.
Even Stanley Gibbons have issued colour charts that have been "updated" by changing colour descriptions and definitions.
Login to Like
this post
re: It is way past time for some software engineer to write a stamp identification piece of software.
I have just started a new thread called "ChromaMate - colour matching software".
I had initially started to post here, but thought that it might be useful to start a new thread instead.
Clive
1 Member
likes this post.
Login to Like.